Initializing Experience...

Portfolio / 2026

Gaurav
Kumar
Nayak

Full Stack Developer & AI Engineer

B.Tech CSE · C.V. Raman Global University · 2024–2028

AI Engineering Intern at FlyRank · Open to AI roles post-2028Remote-friendly
Competed & Won At
IIT Bombay
Kaizen × ARIES × NyneOS Hackathon, Mar 2026
Amazon
Nova AI Hackathon — Prize Winner, $200 AWS Credits
Meta
Meta × OpenEnv — Top 800 / 31,000+ Teams
IISc
ArtPark CodeForge Hackathon 2026

Scroll to explore

01 — Philosophy

I turn
magic into
logic.

AI isn't magic — it's architecture, iteration, and obsession. I just make it look effortless.

02 — Approach

I engineer
what others
imagine.

Where curiosity meets code — designing intelligent systems that don't just work, but make people wonder how.

03 — Selected Work

See what I've built

About Me

Who I Am

3rd-Year CS Student.
AI Systems & Full-Stack Engineer.

Building production-grade systems and agentic workflows.

Amazon Prize

Nova AI Hackathon '26

Top 800 / 31k+

Meta × OpenEnv

12 Built

6 Deployed, 6 Prototypes

I'm Gaurav — a computer science student at C.V. Raman Global University specializing in AI engineering and full-stack development. I focus on building reliable software: multi-agent LLM pipelines, custom RAG architectures, and distributed backends. My work has been awarded at Amazon, Meta, and various IIT hackathons. I'm actively seeking AI engineering internships and remote developer roles where I can build real systems from day one.

Current Research Focus

Active research vs past skills — May 2026

ACTIVE
PyTorch
Python
OpenAI

Mechanistic Interpretability

Understanding what transformer attention heads actually encode. Using CircuitScope + GemmaScope + TransformerLens to run ablation studies on coding-related circuits. Goal: explain why fine-tuned models improve on specific benchmarks.

PyTorchPythonDeep LearningAblation Studies
ACTIVE
HuggingFace
OpenAI
Python

Evaluation-First LLM Development

Building evaluation suites before writing a single model prompt. Combining RAGAS (faithfulness, relevancy), DeepEval (hallucination, toxicity), and custom HumanEval variants for code-specific benchmarks.

HuggingFaceOpenAIModel EvaluationEvals
ACTIVE
LangChain
AWS
Redis

Stateful Multi-Agent Systems

Moving beyond single-turn agents to persistent, stateful workflows using LangGraph. Focus on self-correction, contradiction detection, and graceful failure modes for production deployments.

LangGraphLangChainStateful WorkflowsAgents
RESEARCH
NVIDIA
Python
PyTorch

GPU Optimization & Triton Kernels

Writing custom Triton kernels and fusing layers in SRAM to bypass memory bottlenecks and optimize inference speed on standard GPUs.

TritonCUDA JITGPU KernelsPyTorch

Available for Summer 2026 AI internships and remote junior roles

Currently open to AI engineering internships, remote developer roles, research collaborations, and high-impact full-stack opportunities.

Download My CV

GAURAV KUMAR NAYAK | AI ENGINEER

01 — METRICS

Impact in Numbers

Rigorous engineering benchmarks and competitive milestones validating system quality

HEADLINE METRICS
8x

GPU KERNEL FUSION SPEEDUP

Achieved 8.2x speedup on RMSNorm compared to eager PyTorch implementation by fusing GPU memory accesses in SRAM.

TritonForge · Triton Kernel
0.55 ms

QUALEDGE NPU LATENCY

Delivers 0.55ms latency and 41x speedup executing 100% NPU-native on Hexagon HTP via AIMET 8-stage PTQ quantization.

QualEdge · Snapdragon X Elite
12500 tok/s

CODESAGE VLLM THROUGHPUT

High-throughput serving of QLoRA 4-bit fine-tuned models via PagedAttention with custom MCP server integration.

CodeSage · H100 vLLM Serving
87 %

NEURALVAULT RAG FAITHFULNESS

Achieved premium retrieval fidelity via hybrid BM25 + pgvector Reciprocal Rank Fusion, evaluated via RAGAS framework.

NeuralVault · Neo4j GraphRAG
ENGINEERING PERFORMANCE
6 Agents

RAILMIND AUTONOMOUS DISPATCH

Coordinates a 6-agent LangGraph pipeline with XGBoost RAC conflict prediction and PPO RL dynamic rescheduling.

RailMind · Indian Railways
+ 24 % WMAPE

HYPERFLOW DEMAND FORECAST

Achieved +24.28% WMAPE forecast lift on stockout-censored inventory using Tobit MLE loss with Swiggy MCP agents.

HyperFlow · Tobit MLE Engine
50 %

AGENTSENTRY CACHE SAVINGS

Slashes LLM token costs by 50% using suffix-delta caching and guards execution via AST exploit shielding.

AgentSentry · LLM Middleware
8 Deployed

PRODUCTION SYSTEMS

Engineered stateful multi-agent systems, edge AI quantizers, GPU kernels, and distributed commerce engines.

Featured Star Projects Grid
RECOGNITION & BENCHMARKS
PRIZE WINNER
Amazon
Devpost

Amazon Nova AI Hackathon

Won exclusive Bonus Blog Post Prize ($200 AWS Credits) — the only team to receive this award. Built multi-agent AI research automation.

TOP GLOBAL AWARD
GLOBAL FINALIST
Meta
HuggingFace
Scaler

Meta × OpenEnv Hackathon

Ranked in the Top 800 globally out of 31,000+ teams in a flagship reinforcement learning challenge judged by Meta AI researchers.

TOP 800 WORLDWIDE
CERTIFIED
Stanford
IBM
Meta
IISc

Professional Credentials

15+ High-impact certifications from Stanford, IBM (GenAI Specialist), Meta (Full Stack), and IITs in AI/ML & System Design.

180+ HOURS MASTERY
02 — CODE ACTIVITY

GitHub Contributions

Every day commit heatmap tracking active AI orchestration pipelines and MLOps integrations

Contributions510+
Active Days148
Max Streak10 days
GitHub Stats Dashboard
Total Commits510+
PRs Merged24
Total Issues8
Contributed Repos12
Top Languages
Python48.5%
TypeScript32.1%
JavaScript12.4%
HTML / CSS7.0%
365 Days Commit Trajectory
Less
More
03 — CAPABILITIES

Technical Proficiency

Recruiter-focused capability indexes showing depth and hands-on production expertise

04 — EVOLUTION

Developer Growth Arc

IIT Bombay
2023

Foundation & Problem Solving

Mastered Python, C++, and core data structures. Solved 100+ LeetCode problems and began building automated scripts and CLI workflows.

MarSolved problem #100 on LeetCode — first algorithm that took 3 days
LeetCode Profile
OctBuilt TaskMgr CLI — 200 lines of Python, first real project
GitHub Repo
PythonC++DSABash
Meta
IBM
Stanford
2024

Full Stack Web & Early ML

Transitioned to professional web architectures and predictive machine learning models. Learned state management, API designs, and dataset vectorization.

FebCompleted Meta Full Stack Certificate — 10 courses, 240+ hours
Verify Certificate
JunIBM HR Attrition Predictor — first ML model with 0.93+ AUC on real data
GitHub Repo
SepCineNexus v0.1 — first project actual users tried
GitHub Repo
ReactFastAPIMongoDBPyTorch
IIT Roorkee
2025

Architecture, Caching & Scalability

Focused on high-performance distributed systems, low-latency microservices, relational range partitioning, and advanced indexing strategies.

JanCineNexus production launch — first system handling real concurrent users
AprFirst microservices migration — learned what Kafka is actually for
AugStarted NeuralVault — first time I understood why RAG is hard
PostgreSQLKafkaRedisDocker
Amazon
HuggingFace
IISc
2026Now

Agentic AI & Mechanistic Interpretability

Building production LLM systems. GPU optimization via Triton, mechanistic interpretability causal patching, multi-agent LangGraph flows, and serverless MLOps infrastructure.

JanIBM GenAI Specialist — understood the gap between demos and production LLMs
FebSkillBridge — IISc Bangalore CodeForge winner mapping skills via topological sort
GitHub Repo
AprAmazon Nova Hackathon — first prize, first published technical blog
Published AWS Blog
AprMeta × OpenEnv — Top 800 / 31,000+ globally
Hackathon Entry
JunRailMind — 6-agent LangGraph AI dispatch intelligence for Indian Railways
Live App
JunHyperFlow — Solved 4 documented Swiggy/Zomato ML production gaps; Gemini 2.0 Flash ReAct agent with live Swiggy MCP APIs
Live App
JulCodeSage — QLoRA fine-tuned Llama 3.3 8B on 52K pairs; +7.1pp HumanEval, vLLM at 12,500 tok/s on H100
Live App
JulQualEdge — Full Qualcomm AIMET → AI Hub → Snapdragon X Elite pipeline; 0.55ms NPU latency, 41× speedup
Live App
QLoRAvLLMLangGraphFastAPIPyTorchPPO RLAIMETMCPGemini
FlyRank
July 2026 — PresentNow

AI Engineering Internship at FlyRank

Building automated SEO and content optimization pipelines using multi-agent LLM systems, structuring schema markup for AI-powered search engines, and configuring llms.txt standard files to optimize visibility.

JulStarted AI Engineering Internship — automating optimization pipelines
Confirmation Letter
LangGraphSEOAgent ArchitecturesSchema Markup
CAPABILITIES

What I Build

Systems I've shipped and the problems they solved

Multi-Agent LLM PipelinesAmazon Prize Winner

Multi-Agent LLM Pipelines

I design and ship autonomous agent systems using LangGraph and CrewAI. ResearchPilot orchestrates 5 specialized nodes with persistent state, self-correction, and contradiction detection — delivering cited research reports in under 5 minutes.

LangGraphCrewAIAmazon NovaAWS Bedrock
GPU Kernel & Systems Engineering8.2x Speedup

GPU Kernel & Systems Engineering

I write custom GPU kernels using OpenAI Triton and PyTorch JIT to bypass memory-bandwidth bottlenecks in LLM inference. TritonForge fuses memory-bound activations directly within SRAM, delivering CUDA-equivalent throughput with 99.2% HBM memory savings.

TritonPyTorch JITCUDA JITGPU Programming
Production Full-Stack Systems8-System Architecture

Production Full-Stack Systems

I build distributed platforms with React 19, Next.js, and FastAPI. CineNexus handles global watch parties, real-time WebSocket sync, and AI-driven personalization across an 8-system microservices architecture.

React 19Next.jsFastAPIWebSockets
RAG Architectures & Vector Search87% RAG Faithfulness

RAG Architectures & Vector Search

I implement hybrid retrieval systems combining dense pgvector embeddings with sparse BM25. LexRAG achieved 87% RAGAS faithfulness vs 61% for naive RAG, using GraphRAG over a Neo4j knowledge graph.

GraphRAGNeo4jpgvectorRAGAS
MLOps & Production FinTech ObservabilityTree SHAP Explainability

MLOps & Production FinTech Observability

I build production-grade FinTech observability and intelligence platforms. Artha AI features supervised ensemble fraud classification, real-time Tree SHAP attributions, SEBI compliance circular RAG reranked with cosine similarity, and real-time MLOps data drift auditing.

FastAPIGradioTree SHAPEvidently AI
NeuroScope: Agentic InterpretabilityGemmaScope SAEs · KL Patching

NeuroScope: Agentic Interpretability

I build mechanistic interpretability platforms. NeuroScope captures residual streams, attention patterns, and SAE features at every step of a multi-turn agent, enabling cross-step causal activation patching and representation steering.

GemmaScopeTransformerLensSAEsNext.js
Agent Security & Prompt FirewallsAST Exploit Shield · 50% Cache Savings

Agent Security & Prompt Firewalls

I design secure agent runtimes and optimization middleware. AgentSentry guards remote code execution via recursive AST parsing, intercepts subshell breakouts, and slashes token costs by 50% using suffix-delta caching.

AST ParsersLLM SecurityPrompt CachingDocker
QLoRA Fine-Tuning & LLM Serving+7.1pp HumanEval · 12,500 tok/s

QLoRA Fine-Tuning & LLM Serving

I fine-tune domain-specific code LLMs using QLoRA DoRA NF4 4-bit quantization on 52K instruction pairs (+7.1pp HumanEval). CodeSage serves models via vLLM PagedAttention at 12,500 tok/s on H100 with an MCP server exposing complete, review, and test-generation tools.

QLoRAvLLMModalMCP
Snapdragon Edge AI & AIMET Quantization0.55ms NPU Latency · 41× Speedup

Snapdragon Edge AI & AIMET Quantization

I optimize and deploy NPU-native ML pipelines to edge silicon using Qualcomm AIMET 8-stage PTQ and AI Hub compilation. QualEdge executes 100% NPU-native on Hexagon HTP with 0 CPU fallbacks, delivering 0.55ms latency and 194.62× model compression.

AIMETQualcomm AI HubHexagon HTPFastAPI
Autonomous Dispatch & RL Control Systems6-Agent LangGraph · ECDSA Audit

Autonomous Dispatch & RL Control Systems

I architect mission-critical AI control platforms for large-scale physical networks. RailMind coordinates a 6-agent LangGraph dispatch pipeline for Indian Railways with XGBoost RAC conflict prediction, BFS delay cascade propagation, PPO RL dynamic rescheduling, and an ECDSA cryptographic audit chain.

LangGraphXGBoostPPO RLECDSA
Hyperlocal Commerce Intelligence & Production ML+24.28% WMAPE · Swiggy MCP Agent

Hyperlocal Commerce Intelligence & Production ML

I solve documented production ML gaps in supply chain and e-commerce. HyperFlow handles stockout-censored demand via Tobit MLE (+24.28% WMAPE lift), velocity-normalized RF gates suppressing ETA display jitter by 81.4%, and Gemini 2.0 Flash ReAct agents with live Swiggy MCP APIs.

Tobit MLELightGBMGemini 2.0Swiggy MCP
ELITE

Global Recognition & Ecosystem

Recognized At

Competitions, certifications, and programs hosted by India's leading institutions and global tech companies

Amazon
Amazon
Nova AI Hackathon — Prize Winner
Meta
Meta
Meta × OpenEnv — Top 800 / 31,000+
HuggingFace
HuggingFace
Meta × OpenEnv — Top 800 / 31,000+
IIT Bombay
IIT Bombay
Kaizen × ARIES × NyneOS Hackathon, Mar 2026
IISc
IISc
ArtPark CodeForge Hackathon 2026
IBM
IBM
IBM GenAI Specialist & Data Science Professional Certification
Stanford
Stanford
Stanford Machine Learning Specialization
IIT Delhi
IIT Delhi
Tryst'26 IIT Delhi Finalist
IIT Roorkee
IIT Roorkee
Zaggle × COMET'26 IIT Roorkee Finalist
Amazon
Amazon
Nova AI Hackathon — Prize Winner
Meta
Meta
Meta × OpenEnv — Top 800 / 31,000+
HuggingFace
HuggingFace
Meta × OpenEnv — Top 800 / 31,000+
IIT Bombay
IIT Bombay
Kaizen × ARIES × NyneOS Hackathon, Mar 2026
IISc
IISc
ArtPark CodeForge Hackathon 2026
IBM
IBM
IBM GenAI Specialist & Data Science Professional Certification
Stanford
Stanford
Stanford Machine Learning Specialization
IIT Delhi
IIT Delhi
Tryst'26 IIT Delhi Finalist
IIT Roorkee
IIT Roorkee
Zaggle × COMET'26 IIT Roorkee Finalist
Python
Python
PyTorch
PyTorch
LangChain
LangChain
OpenAI
OpenAI
Anthropic
Anthropic
Docker
Docker
Kubernetes
Kubernetes
FastAPI
FastAPI
React
React
Next.js
Next.js
MongoDB
MongoDB
PostgreSQL
PostgreSQL
Python
Python
PyTorch
PyTorch
LangChain
LangChain
OpenAI
OpenAI
Anthropic
Anthropic
Docker
Docker
Kubernetes
Kubernetes
FastAPI
FastAPI
React
React
Next.js
Next.js
MongoDB
MongoDB
PostgreSQL
PostgreSQL
02 — Engineering Portfolio

Featured Projects

Selected systems engineering and AI research projects benchmarked on production performance metrics.

ResearchPilot
Amazon

ResearchPilot

Amazon Prize Winner · Multi-Agent LLM Orchestrator

The Problem

Manual research synthesis takes 4–6 hours of tab-switching and cross-referencing across dozens of sources.

What I Built

5-node LangGraph pipeline with Search, Read, Reflect, Cross-Check, and Write agents. Self-corrects via reflection loops. Cites every claim.

Impact

Amazon Prize WinnerAWS Blog Post PublishedSub-5min reports

Primary Tech Stack

Amazon Nova Models (Micro, Lite, Pro, Premier, Canvas)
LangGraph (Stateful workflow orchestration)
AWS Bedrock API (Zero-latency model invocation)
FAISS (Dense vector semantic search & retrieval)
Streamlit (Real-time analyst interactive UI)
RailMind
FastAPI

RailMind

Autonomous Dispatch Intelligence · Indian Railways

The Problem

Indian Railways dispatch runs on fragmented manual coordination — conflict resolution, delay cascade tracking, and rescheduling are siloed with no unified AI decision layer.

What I Built

6-agent LangGraph orchestration pipeline with XGBoost RAC (Route Acquisition Conflict) prediction, BFS cascade propagation for delay chain simulation, PPO reinforcement learning for dynamic rescheduling, and an ECDSA cryptographic audit ledger for tamper-proof decision logging.

Impact

6-Agent LangGraphXGBoost Conflict PredictionECDSA Audit Chain

Primary Tech Stack

LangGraph (Stateful 6-agent dispatch orchestration)
XGBoost (Route Acquisition Conflict prediction)
BFS Cascade Propagation (Delay chain simulation engine)
PPO RL (Dynamic rescheduling reward optimization)
FastAPI (Async REST + SSE backend)
ECDSA (Cryptographic tamper-proof audit ledger)
HyperFlow
Google Gemini

HyperFlow

Hyperlocal Commerce Intelligence · 4 ML Problems Solved

The Problem

Swiggy Bytes and Zomato Engineering document four unsolved production ML problems: censored demand from stockouts, ETA display jitter from GPS noise, cancelled order resale arbitrage, and refund loop fraud from cloud-kitchen geo-collision. No existing platform solves all four.

What I Built

Production ML platform solving each gap from first principles: Heteroscedastic Tobit MLE + LightGBM Quantile for censored demand (+24.28% WMAPE), velocity-normalized RF Classifier gate for ETA jitter (81.4% suppressed), Thermal SQI + Sybil proximity guard for arbitrage (100% blocked), tenure-gated bypass + semantic plausibility engine for fraud (0% false positives). Gemini 2.0 Flash ReAct agent with live Swiggy MCP APIs. Atomic dual-mode locking (Redis Redlock + PostgreSQL FOR UPDATE NOWAIT) at 1,598 req/sec.

Impact

+24.28% WMAPE Lift81.4% ETA Jitter Suppressed100% Arbitrage Blocked1,598 req/sec

Primary Tech Stack

Heteroscedastic Tobit MLE + LightGBM (Censored demand forecasting)
RandomForest Classifier (Velocity-normalized ETA jitter gate)
Cox Proportional Hazards (Dark store profitability survival model)
Gemini 2.0 Flash + LangGraph (ReAct commerce agent with Swiggy MCP)
FastAPI + Redis Redlock + PostgreSQL (Dual-mode atomic inventory locking)
PSI Drift Detection (Auto-retraining trigger every 15s)
CodeSageZ
PyTorch

CodeSageZ

Graph-Augmented RAG · Tree-sitter AST · QLoRA 4-bit

The Problem

General-purpose code LLMs underperform on domain-specific repository completion and review tasks, and naive vector search misses multi-hop structural call dependencies.

What I Built

Built a Retrieval-Augmented Generation (RAG) engine that expands vector seeds via 1-hop BFS over a Tree-sitter AST call graph and re-ranks with a hybrid score (0.6×cosine + 0.4×graph proximity). Fine-tuned Qwen2.5-Coder-1.5B via 4-bit QLoRA (Unsloth) on 8K CommitPack bug-fix commits.

Impact

0% → 53.3% Recall@8+9.38 CodeBLEU pts (+15.5%)p95 5.8 ms Retrieval

Primary Tech Stack

FastAPI + Next.js 14 (High-performance API & interactive UI)
Tree-sitter + NetworkX (AST parsing & 1-hop BFS call graph expansion)
QLoRA + Unsloth (4-bit Qwen2.5-Coder-1.5B fine-tuning on 3.8 GB VRAM)
ChromaDB + PostgreSQL (Hybrid dense vector & graph metadata store)
Docker + pytest (Containerised deployment & automated testing)
QualEdge
Python

QualEdge

Qualcomm Snapdragon Edge AI · 0.55ms NPU · 41× Speedup

The Problem

No production-grade open platform demonstrates the full Qualcomm ML stack end-to-end — from AIMET quantization through AI Hub compile and real Snapdragon NPU profiling — with reproducible, sourced benchmarks.

What I Built

Full 8-stage AIMET PTQ & AI Hub compilation pipeline (BN fold, CLE, ReLU surgery, AdaRound INT8, QNN compile job j5w110q4g, HTP profile job jgdzzyo65). Achieves 100% Hexagon HTP-native execution with 0 CPU fallback operators. Hybrid on-device/cloud LLM router (ModernBERT + LogReg) with 1.85ms p95 router latency and 100% local HTP cloud avoidance.

Impact

0.55ms NPU Latency41× Speedup194.62× Compression100% Hexagon HTP Native

Primary Tech Stack

AIMET Quantization Suite (BN fold, CLE, ReLU surgery, AdaRound INT8)
Qualcomm AI Hub (Real Snapdragon X Elite CRD compile & profile)
Hexagon Tensor Processor (100% HTP-native, 0 CPU fallback operators)
ModernBERT + LogReg (Hybrid on-device/cloud router classifier)
FastAPI + React 19 (Live silicon telemetry console dashboard)
CineNexus
Next.js

CineNexus

8-System Distributed Watch Parties Portal

The Problem

Friends watching movies together across time zones have no synchronized, social streaming experience with personalization.

What I Built

8-service distributed platform: Next.js frontend, FastAPI backend, WebSocket sync engine, Redis pub/sub, AI recommendation layer, and PostgreSQL.

Impact

Real-time sync across global usersSub-50ms WebSocket latencyAI-personalized recommendations

Primary Tech Stack

React 19 & Next.js 15 (Streaming SSR layout)
FastAPI (Asynchronous high-throughput gateway)
WebSockets (Dynamic real-time synchronization)
Redis Cache (Upstream rate limiting and session store)
PostgreSQL (Relational user metadata warehouse)
TritonForge
NVIDIA

TritonForge

GPU Kernel Engineering · OpenAI Triton (Up to 8.2× Faster)

The Problem

Memory-bound LLM bottlenecks (like SwiGLU activations and RMSNorm) limit throughput and waste high-bandwidth memory (HBM) on modern GPUs.

What I Built

A suite of fused CUDA-equivalent GPU kernels written in OpenAI Triton including SwiGLU, RMSNorm, and FlashAttention-2. Fuses memory-bound operations directly within GPU SRAM, breaking the memory wall to deliver up to 8.2× speedup over PyTorch eager mode.

Impact

8.2× RMSNorm Speedup99.2% HBM Saved91% Peak BW300+ Stars

Primary Tech Stack

OpenAI Triton 3.0+ JIT compiler
PyTorch 2.4+ (Custom autograd extensions)
CUDA 12.3 (Hardware target execution context)
NVIDIA A100 GPUs (Profile target architecture)
Python 3.9+ (Custom kernel test suites)
Artha AI
FastAPI

Artha AI

Tree SHAP Explainability · RegGuard RAG Platform

The Problem

Financial ensemble models predict fraud with high accuracy, but compliance auditors cannot act on them without explainable features and regulatory alignment.

What I Built

Voting ensemble (XGBoost + LightGBM + Random Forest) + real-time Tree SHAP values + pgvector compliance document matching RAG and LangChain SQL agents.

Impact

Tree SHAP Attribution Latency <20ms99.1% Fraud Match PrecisionSEBI & NPCI Circular Auditing

Primary Tech Stack

FastAPI (Asynchronous high-throughput gateway)
PostgreSQL & pgvector (Transactional & vector store)
LangChain (Adaptive audit database agent)
SHAP (Real-time Tree SHAP local feature attributions)
Evidently AI (MLOps target and data drift monitoring)
More Work

Also Notable Work

Other shipped systems, ML pipelines, and competitive hackathon builds

AI/ML
6-Agent LangGraph · ECDSA Audit Chain

6-agent LangGraph AI dispatch intelligence for Indian Railways. Features XGBoost RAC conflict prediction, BFS delay cascade propagation, PPO dynamic rescheduling, and an ECDSA cryptographic audit ledger.

LangGraph
LangGraph
XGBoost
XGBoost
FastAPI
FastAPI
PPO RL
PPO RL
ECDSA
ECDSA
FastAPI
Python
AI/ML
+24.28% WMAPE · Swiggy MCP Agent

Hyperlocal commerce intelligence platform solving 4 documented Swiggy/Zomato ML gaps: Tobit MLE demand (+24.28% WMAPE), RF ETA jitter gate (81.4% suppressed), SQI arbitrage shield, and Gemini 2.0 Flash ReAct agent with live Swiggy MCP.

FastAPI
FastAPI
LightGBM
LightGBM
Google Gemini
Gemini
Redis
Redis
LangGraph
LangGraph
FastAPI
Redis
Python
RESEARCH
+7.1pp HumanEval · 12,500 tok/s vLLM

QLoRA fine-tuned Llama 3.3 8B model on 52K code pairs (+7.1pp HumanEval). Served via vLLM PagedAttention at 12,500 tok/s on H100 with a 4-tool MCP server for Claude Desktop and Cursor.

PyTorch
PyTorch
vLLM
vLLM
Modal
Modal
HuggingFace
HuggingFace
MCP
MCP
PyTorch
HuggingFace
Python
RESEARCH
0.55ms NPU Latency · 41× Speedup

Full Qualcomm AIMET PTQ pipeline targeting Snapdragon X Elite NPU with 0.55ms latency (41× speedup). Features hybrid on-device/cloud LLM router classifier at 93.3% accuracy and real-time telemetry.

AIMET
AIMET
Qualcomm AI Hub
Qualcomm AI Hub
Python
Python
ModernBERT
ModernBERT
FastAPI
FastAPI
Python
FastAPI
RESEARCH
3.2x avg speedup across fused kernels

Memory-bound LLM bottleneck solver using OpenAI Triton. Fuses memory-bound activation layers and RMSNorm directly within GPU SRAM, eliminating HBM read/write overhead to deliver CUDA-equivalent performance.

Python
Python
PyTorch JIT
PyTorch JIT
Triton JIT
Triton JIT
CUDA JIT
CUDA JIT
Python
PyTorch
NVIDIA
FULL STACK
Tree SHAP Explainability · RegGuard RAG

FinTech audit platform featuring ensemble transaction fraud matching, real-time Tree SHAP explainability, online SEBI/NPCI circular RAG, and LangChain SQL statement auditing.

FastAPI
FastAPI
PostgreSQL
PostgreSQL
LangChain
LangChain
Redis
Redis
FastAPI
Redis
PostgreSQL
Python
RESEARCH
GemmaScope SAEs · Trajectory KL Patching

Mechanistic interpretability platform mapping LLM agent trajectories. Decomposes residual streams using 16k-width GemmaScope Sparse Autoencoders and runs causal activation patching to isolate reasoning failures.

PyTorch
PyTorch
Python
Python
OpenAI
OpenAI
React
React
PyTorch
Python
OpenAI
React
AI/ML
87% Faithfulness · Neo4j GraphRAG

A legal AI research assistant utilizing dense pgvector embeddings with sparse BM25 and Neo4j GraphRAG. Achieves 87% faithfulness, evaluated using the RAGAS framework across five custom, modular RAG pipelines.

LangGraph
LangGraph
LangChain
LangChain
PostgreSQL
PostgreSQL
AWS
AWS
Python
Python
PostgreSQL
AWS
Python
OpenAI
AI/ML
Top 800 Meta · OpenEnv RL challenge

A stateful multi-agent customer support ticket triage and routing platform built for Meta × OpenEnv. Spawns specialized CrewAI solvers in a LangGraph pipeline to route and solve tickets dynamically.

LangGraph
LangGraph
CrewAI
CrewAI
FastAPI
FastAPI
Redis
Redis
HuggingFace
HuggingFace
FastAPI
Redis
HuggingFace
Python
FULL STACK
8-System Distributed · Playback WebSockets

A distributed movie recommendation and discovery system handling watch parties, real-time playback sync, and AI-driven personalization across an 8-system microservices architecture.

Next.js
Next.js
React
React
FastAPI
FastAPI
Redis
Redis
PostgreSQL
PostgreSQL
Next.js
React
FastAPI
Redis
PostgreSQL
HACKATHON
Finalist · Devcation Delhi 2026

AI-driven women's safety platform providing real-time threat detection via Whisper + BERT ambient voice monitoring, one-tap emergency SOS calls, location tracking, and visual safety heatmaps.

React
React
FastAPI
FastAPI
HuggingFace
HuggingFace
MongoDB
MongoDB
Python
Python
React
Node.js
FastAPI
HuggingFace
MongoDB
Python
HACKATHON
Winner · COMET'26 IIT Roorkee

Multi-agent financial operating system running CrewAI solvers to execute real-time spend intelligence, 13-week cash flow forecasts, month-end reconcile loops, and transaction auditing.

CrewAI
CrewAI
LangChain
LangChain
OpenAI
OpenAI
PostgreSQL
PostgreSQL
Redis
Redis
OpenAI
FastAPI
PostgreSQL
Redis
HACKATHON
Winner · CodeForge IISc Bangalore

AI-adaptive onboarding engine that parses resumes + JDs via Gemini, maps skills gaps to the O*NET taxonomy, and dynamically generates prerequisite-ordered pathways via Kahn's topological sort.

FastAPI
FastAPI
React
React
Docker
Docker
Python
Python
Next.js
Next.js
FastAPI
React
Docker
Python
Next.js
HACKATHON
Finalist · HackMatrix 2.0 IIT Patna

Bilingual ambient medical AI scribe for Indian clinics. Converts Hindi + English doctor-patient consultations into FHIR R4-compliant clinical summaries in under 2.5 seconds using BioBERT NER.

FastAPI
FastAPI
Claude AI
Claude AI
PostgreSQL
PostgreSQL
Python
Python
React
React
FastAPI
Claude AI
PostgreSQL
Python
React
HACKATHON
Finalist · Technex'26 IIT BHU

Real-time AI avatar interview system. Integrates custom D-ID avatars, Whisper audio transcribers, MediaPipe facial markers analysis, and XGBoost models for multi-dimensional grading.

FastAPI
FastAPI
React
React
Claude AI
Claude AI
Python
Python
Next.js
Next.js
FastAPI
React
Claude AI
Python
Next.js
RESEARCH
Kaizen Winner · IIT Bombay

ML-driven GNSS spoofing detector utilizing hybrid LSTM Autoencoder + XGBoost models. Processes 18 high-frequency features on TEXBAT signals with 97%+ recall and under 15ms latency.

Python
Python
PyTorch
PyTorch
FastAPI
FastAPI
Docker
Docker
AWS
AWS
Python
PyTorch
FastAPI
Docker
AWS
RESEARCH
AST Shell Exploit Shield · 50% Cache Savings

Security science and prompt caching middleware for autonomous agents. Intercepts AST shell exploit attempts and reduces LLM API costs by 50% using suffix-delta caching.

Python
Python
AST Parsers
AST Parsers
Docker
Docker
Prompt Hashing
Prompt Hashing
Security Auditing
Security Auditing
Python
Docker
Guardrails AI
RECOGNITION

Hackathons & Achievements

Row 1 — Star Achievements
Amazon Nova AI Hackathon preview
PRIZE WINNERApr 2026

Amazon Nova AI Hackathon

Amazon Web Services · Devpost

Project Shipped

ResearchPilot (Multi-Agent Research Pipeline)

The Problem

Manual research synthesis takes 4–6 hours of tab-switching and cross-referencing across dozens of sources.

What I Built

5-node LangGraph pipeline with Search, Read, Reflect, Cross-Check, and Write agents. Self-corrects via reflection loops. Cites every claim.

Impact

Amazon Prize Winner · AWS Blog Post Published · Sub-5min reports

Amazon
Devpost
GitHub
Meta × OpenEnv Global AI Agent Hackathon preview
GLOBAL FINALISTApr 2026

Meta × OpenEnv Global AI Agent Hackathon

Meta AI · HuggingFace · Scaler

Project Shipped

SupportOps-Env (RL environment for AI agents)

The Problem

Customer support at scale requires consistent, policy-compliant responses but human agents produce inconsistent quality under volume.

What I Built

Multi-agent support automation: intent classifier, policy retrieval (RAG), response generator, quality checker, and escalation router.

Impact

Top 800 / 31,000+ globally — Meta × HuggingFace Challenge

Meta
HuggingFace
Scaler
GitHub
ArtPark CodeForge Hackathon preview
WINNERFeb 2026

ArtPark CodeForge Hackathon

IISc Bangalore

Project Shipped

SkillBridge (Gemini topological sort onboarding)

The Problem

Corporate onboarding programs offer generic reading tracks rather than customized, skill-gap aligned pathways.

What I Built

SkillBridge corporate onboarding engine: maps JD gap analyses via Gemini and plots learning paths via Kahn's topological sort.

Impact

Winner — ArtPark CodeForge IISc Bangalore

IISc
GitHub
Row 2 — National & Regional Finalists
Economic Times AI Hackathon preview
SEMI-FINALISTMar 2026

Economic Times AI Hackathon

The Economic Times

Project Shipped

MarketMind (Multi-Agent Market Analyst)

The Problem

Analyzing NSE & BSE market opportunities requires synthesis of massive data streams (news, filings, indicators, and holdings) under intense time pressure, which takes hours for retail investors.

What I Built

Multi-agent AI analyst console coordinating 4 specialized agents (Market Data, Technicals, ET News/BSE Filings, and Portfolio-aware) to synthesize stock investment responses in under 30 seconds.

Impact

Semi-Finalist (Problem Statement 6) — Economic Times GenAI Hackathon

Economic Times
GitHub
Neural Nexus IIT Jammu preview
FINALISTApr 2026

Neural Nexus IIT Jammu

IIT Jammu

Project Shipped

FloatChat (ARGO RAG Oceanographic Widget)

The Problem

The ARGO program generates terabytes of complex oceanographic data that usually requires specialized tools and expert domain knowledge to interpret, blocking general accessibility.

What I Built

FloatChat oceanographic assistant: semantic FAISS retrieval (all-MiniLM-L6-v2) for top profiles, Groq LLaMA-3 natural language queries, and 6 interactive auto-triggered chart types (SST time series, depth profiles, T-S diagrams).

Impact

Round 1 Finalist — Neural Nexus, IIT Jammu · Ocean data without domain expertise

IIT Jammu
GitHub
DevFusion IIT Bombay × IIT Delhi preview
FINALISTApr 2026

DevFusion IIT Bombay × IIT Delhi

IIT Bombay · IIT Delhi

Project Shipped

PrepGrid (AI Interview Platform)

The Problem

Off-the-shelf mock interview tools offer static question prompts and lack dynamic domain evaluation feedback.

What I Built

PrepGrid mock interview platform: real-time voice-to-text with Whisper, topological skills scoring, Next.js SSR, and FastAPI backend sockets.

Impact

DevFusion Finalist · 200+ mock scenarios handled during active trials

IIT Bombay
IIT Delhi
GitHub
HackMatrix 2.0 IIT Patna preview
FINALISTApr 2026

HackMatrix 2.0 IIT Patna

IIT Patna

Project Shipped

VaidyaScribe (FHIR Ambient Scribe)

The Problem

Rural Indian clinics require fast, bilingual medical scribing, but manual record entry takes hours and causes clinical reporting delays.

What I Built

Mobile-first VaidyaScribe dictation: transcribes Hindi/English Whisper dialogues, indexes BioBERT NER, and formats FHIR R4 reports.

Impact

Finalist — HackMatrix 2.0 IIT Patna

Devcation Delhi 2026 preview
FINALISTJan 2026

Devcation Delhi 2026

Devcation Delhi

Project Shipped

ShieldHer (Ambient Safety System)

The Problem

Passive ambient tracking is rarely integrated with audio triggers, delaying emergency alarms during sudden voice-only threat conditions.

What I Built

Passive ambient threat monitor processing speech to text via Whisper, classifying distress intent using BERT, and triggering location-linked alerts.

Impact

Finalist — Devcation Delhi 2026

Technex'26 IIT BHU preview
FINALISTMar 2026

Technex'26 IIT BHU

IIT BHU

Project Shipped

InterviewX (AI Avatar Evaluator)

The Problem

Traditional technical screening interviews lack standardized bias-free grading and real-time behavioral insights.

What I Built

Real-time AI avatar interview system with D-ID avatars, Whisper audio transcription, and MediaPipe facial markers.

Impact

Finalist — Technex'26 IIT BHU

Row 3 — Verified Participations
PARTICIPANTMar 2026

IIT Bombay Kaizen × ARIES × NyneOS

IIT Bombay

Project Shipped

GNSS Anti-Spoofing (LSTM Security)

ML-driven GNSS anti-spoofing pipeline using a hybrid LSTM Autoencoder and XGBoost. Processed 18 high-frequency features on TEXBAT signal datasets.

PARTICIPANTMar 2026

Zaggle × COMET'26 IIT Roorkee

IIT Roorkee · Zaggle

Project Shipped

CFO-OS (Multi-Agent OS)

Multi-agent spend intelligence: autonomous cash flow forecasting, 13-week cash projections, and double-entry transaction auditing.

IIT Roorkee
Zaggle
GitHub
CREDENTIALS

Courses & Certifications

Featured Specializations & Professional Programs

SPECIALIZATION · 10 Courses · 240+ Hours
Meta
Coursera

Meta Full Stack Developer Specialization (10-course)

Meta's official 10-course full stack program used internally at Meta engineering teams. Covers React, Django, MySQL, REST APIs, version control workflows, and UX principles. 240+ hours — directly applied to CineNexus and PrepGrid.

Key Skills

ReactDjangoMySQLREST APIsGitUX Principles
Issued: Apr 2026
Verify
SPECIALIZATION · 3 Courses · 100+ Hours
Stanford
DeepLearning.AI

Machine Learning Specialization

Andrew Ng's flagship 3-course ML program covering supervised learning (linear/logistic regression, SVMs), unsupervised learning (clustering, anomaly detection), recommender systems, and reinforcement learning. The gold standard ML certification — 100+ hours.

Key Skills

Supervised LearningUnsupervised LearningNeural NetworksRecommender SystemsReinforcement Learning
Issued: Apr 2026
Verify
SPECIALIZATION · 10 Courses · 200+ Hours
IBM
Coursera

IBM Data Science Professional Certificate

10-course end-to-end data science program: Python, SQL, data visualization with Matplotlib and Seaborn, ML with Scikit-learn, and model deployment on IBM Cloud. Capstone: IBM HR Attrition Prediction achieving 0.93+ AUC.

Key Skills

PythonSQLMatplotlibSeabornscikit-learnModel Deployment
Issued: Apr 2026
Verify
SPECIALIZATION · 7 Courses · 80+ Hours
IBM
Coursera

Generative AI Engineering with LLMs (7-course)

7-course IBM program covering LLM architecture, RAG pipelines, fine-tuning with PEFT/LoRA, prompt engineering, LangChain agent development, and deploying AI applications with IBM Watson. Directly applied to CircuitScope and NeuralVault projects.

Key Skills

LLMsRAGFine-tuningPEFT/LoRAPrompt EngineeringLangChain
Issued: Apr 2026
Verify
SPECIALIZATION · 12 Courses · 240+ Hours
IBM
Coursera

IBM Full Stack Software Developer Specialization (12-course)

12-course professional certificate covering cloud-native full stack development with React, Node.js, Python/Flask, Django, REST APIs, containers, Kubernetes, microservices, and CI/CD pipelines. Capstone deployed on IBM Cloud.

Key Skills

ReactNode.jsDockerKubernetesDjangoFlaskCI/CD
Issued: Apr 2026
Verify

Targeted Courses & Skill Certifications

AWS
Udemy
Mar 2026

Ultimate AWS Certified Cloud Practitioner CLF-C02

Udemy · 20 Hours

Complete AWS Cloud Foundations prep by Stéphane Maarek covering core AWS services (EC2, S3, Lambda, RDS, IAM), cloud architecture principles, security, pricing models, and the Well-Architected Framework. Applied to NeuralVault and ResearchPilot deployments.

AWS
Verify
IIT Guwahati
Coursera
Mar 2026

Programming with Generative AI

IIT Guwahati × Coursera · 24 Hours

IIT Guwahati's GenAI program covering generative AI fundamentals, prompt engineering patterns, LLM architecture internals, and building production AI applications with IBM Watson and open-source models.

GenAI
Verify
Stanford
DeepLearning.AI
Apr 2026

Advanced Learning Algorithms

Stanford × DeepLearning.AI · 30 Hours

Deep dive into neural networks built from scratch in NumPy and TensorFlow, decision trees, random forests, XGBoost, and practical ML advice on diagnosing bias/variance and improving model performance. Course 2 of the ML Specialization.

Neural Networks
Verify
09 // PROVEN IMPACT

Verified Recognition

Verifiable endorsements of architectural execution, AI performance optimizations, and technical design from peer reviews and global hackathons.

Building the IBM HR Attrition Prediction System was a masterclass in data-driven problem solving. We engineered complex ML models and interactive dashboards that unlocked real-world insights.

Ankit Dash
IBM

Ankit Dash

Project Collaborator · C.V. Raman Global University

ResearchPilot won the Bonus Blog Post Prize for its deep technical breakdown of multi-agent LLM orchestration using Bedrock. Gaurav's system is highly detailed, handling state, self-correction, and contradiction detection.

AE
Amazon

AWS Editorial Team

Official Bonus Blog Post Prize · Amazon Web Services

10 — WRITING

Writing & Technical Case Studies

Technical deep-dives into LLM engineering, interpretability research, and full-stack architecture.

8 min read
April 2026
AWS Builder

ResearchPilot: Turning Hours of Research into Minutes with Amazon Nova

Exploring the multi-agent orchestration of Nova Micro, Lite, and Pro for high-speed cross-source research and contradiction detection. Won the Bonus Blog Post Prize at Amazon Nova AI Hackathon.

LLM Orchestration Amazon Nova Multi-Agent
12 min read
May 2026
Case Study

TritonForge: Fusing SwiGLU and RMSNorm Kernels in OpenAI Triton

A deep dive into custom GPU memory alignment, SRAM tile optimization, and reducing HBM read/write roundtrips for LLM activation layers. Achieved 3.2x average speedup.

GPU Kernels OpenAI Triton CUDA Optimization
10 min read
March 2026
Case Study

NeuralVault: Building a GraphRAG Retrieval Pipeline on Neo4j & pgvector

Architecting hybrid search (dense pgvector + sparse BM25) and connecting multi-hop entity graphs in Neo4j to boost RAG faithfulness to 87%, evaluated via RAGAS.

GraphRAG Neo4j pgvector RAGAS Evaluation
15 min read
June 2026
Case Study

Artha AI: Explainable Machine Learning & Regulatory RAG in Production FinTech Systems

Designing secure financial transaction pipelines with ensemble fraud voting classifiers, Tree SHAP attributions under 20ms, and pgvector compliance document matching.

Explainable AI FinTech RAG Tree SHAP MLOps
Contact

Got an idea? Let's build it.

Open to AI engineering internships, remote developer roles, and technical collaborations.

Or reach me directly at gauravnayak711@gmail.com