Hey, I'm Lenin Jacob Regi

CRAFTING INTELLIGENT SOLUTIONS THAT TRANSFORM POSSIBILITIES INTO REALITY.

I'M AN AI ENGINEER DEDICATED TO BUILDING THE FUTURE, ONE ALGORITHM AT A TIME.

GET IN TOUCH

ABOUT ME

Lenin Jacob Regi

I am Lenin Jacob Regi, an AI/ML Engineer, LLM & RAG Specialist, and Full-Stack AI Developer pursuing B.Tech in Computer Science & Engineering (AI/ML) at Karunya Institute of Technology and Sciences, Coimbatore (2023–2027). I have independently built 30+ production AI systems spanning LLMs, RAG pipelines, multi-agent architectures, computer vision, and full-stack applications.

I architect hybrid RAG systems with multi-DB backends (Milvus, pgvector, Neo4j, Elasticsearch), integrate 100+ LLMs via LiteLLM, and optimize inference for 70B+ models on 4GB GPU via layer-wise loading. I also build multi-agent LangGraph systems with real-time full-stack apps using React/Next.js, FastAPI, and Docker.

As Gen-AI Head at Karunya Innovation & Design Studio (KIDS), I lead LLM/SLM research, agentic AI workshops, and hackathon mentorship for 100+ students. I built Karunya-SLM from scratch using GPT-2 architecture, custom BPE tokenizer, and PyTorch training.

DOWNLOAD RESUME

EDUCATION

B.Tech – Computer Science & Engineering (AI/ML)

Karunya Institute of Technology and Sciences, Coimbatore

CGPA: 6.0

2023–2027

Senior Secondary

State Board

Percentage: 100%

2023

Secondary

State Board

Percentage: 74%

2021

EXPERIENCE

AI/ML Engineer – Independent Developer & Open-Source Contributor

2023 – Present

30+ production AI systems: LLMs, RAG pipelines, multi-agent architectures, computer vision, full-stack apps

  • Architected hybrid RAG systems (semantic + FTS + RRF) with multi-DB backends (Milvus, pgvector, Neo4j, Elasticsearch); integrated 100+ LLMs across 7 providers via LiteLLM abstraction.
  • Optimized LLM inference to run 70B+ models on 4GB GPU via layer-wise loading & block quantization — 3× speedup; achieved 100% defect detection (CV) and 93.3% efficiency (Google Solution Challenge).
  • Built multi-agent LangGraph systems with tool use, sub-agent delegation; engineered real-time full-stack apps with React/Next.js, FastAPI/Node.js, WebSocket, and containerized GPU-enabled Docker deployments.

Gen-AI Head – Karunya Innovation & Design Studio (KIDS)

Jan 2025 – Present

Leading LLM/SLM research, agentic AI workshops, and hackathon teams for 100+ students

  • Built and deployed Karunya-SLM from scratch — GPT-2 architecture, custom BPE tokenizer, PyTorch training, WandB tracking.

MY SERVICES

Leveraging expertise in AI/ML/DL, I offer services in intelligent system development, web solutions, and cybersecurity.

AI/ML/DL Development

Building intelligent systems and models using cutting-edge Artificial Intelligence, Machine Learning, and Deep Learning techniques.

Web Design & Development

Crafting visually appealing and user-friendly web applications with robust backend solutions.

Cybersecurity Solutions

Implementing secure practices and developing solutions to protect digital assets and information.

Cryptography & Security

Exploring and applying cryptographic principles for data privacy and secure communication.

GET IN TOUCH

FEATURED PROJECTS

SurfSense – AI Research Agent & Knowledge Management

RAG • LangGraph • LiteLLM • FastAPI • Next.js

20+ source integrations (Notion, Slack, GitHub, YouTube); Deep Agent with 11 tools, 6000+ embedding models, RBAC. 2-tier hierarchical RAG with RRF, Pinecone/Cohere rerankers, podcast generation in <20 sec.

Stack: FastAPI, Next.js, TypeScript, PostgreSQL+pgvector, LangChain, LangGraph, LiteLLM, Redis, Celery, Docker

VIEW ON GITHUB

A.E.G.I.S – AI Code Vulnerability Detection & Auto-Remediation

Security • SAST/DAST • LLM Code Review • Docker

Pre-deployment security platform: SAST (Semgrep, CodeQL) + DAST + LLM code review in zero-trust Kata Containers. CVSS-scored prioritization, automated multi-candidate fix generation, human-in-the-loop approval UI with risk heatmaps.

Stack: Python, Mistral 7B, CodeLlama 13B, Semgrep, CodeQL, Flask, React, Docker

VIEW ON GITHUB

Scoratis – Agentic 3D Socratic Learning Platform

Three.js • LangGraph RAG • Manim • FastAPI

Three.js 3D museum + LangGraph RAG tutor: 11 tools, 5 sub-agents, 6-step Socratic scaffolding, 10 subject domains. 50+ API endpoints, short/long-term memory, Manim animations, multi-LLM with VRAM-based model selection.

Stack: React 18, Three.js, FastAPI, PostgreSQL+pgvector, LangGraph, LiteLLM, Manim, Docker

VIEW ON GITHUB

QUERIX.AI – YouTube Semantic Chatbot with RAG

Whisper • Milvus • Ollama • deepseek-r1

Video pipeline: yt-dlp → Whisper → chapter-aware chunking → Milvus IVF_FLAT → deepseek-r1 Q&A. LLM-generated fallback chapters; 4096-char chunk storage with etcd/MinIO Milvus stack.

Stack: Python, Whisper, yt-dlp, FFmpeg, Pydub, Milvus, Ollama (mxbai-embed-large, deepseek-r1), Docker

VIEW ON GITHUB

maestro-studio – 100% Offline AI Video Generator

llama.cpp • Manim • Piper TTS • FFmpeg • CUDA

LLM script → Manim animation → Piper TTS narration → FFmpeg compositing; zero cloud, full GPU acceleration. Generates 1080p 2–5 min videos in 2–5 min; CLI batch mode, multiple voice personas, cinematic Manim templates.

Stack: Python, llama.cpp (Llama 3.2 3B), Manim, Piper TTS, FFmpeg, Gradio, CUDA

VIEW ON GITHUB

YATHRA – Real-Time Full-Stack Travel Companion

Node.js • Socket.IO • React • MongoDB • Cloudinary

Trip planning, smart expense splitting, memory wall (Cloudinary CDN), real-time group chat via WebSockets. Railway + MongoDB Atlas deploy; Helmet, rate limiting (100 req/15 min), settlement optimization.

Stack: Node.js, Express, MongoDB, Socket.IO, React, TailwindCSS, Cloudinary, Passport.js (OAuth 2.0), Chart.js

VIEW ON GITHUB

TECHNICAL SKILLS

Languages

Python JavaScript TypeScript SQL C++ Scala

AI / ML

PyTorch TensorFlow LangChain LangGraph LiteLLM Hugging Face RAG Milvus pgvector Whisper OpenCV XGBoost WandB

Backend & Infrastructure

FastAPI Flask Django Node.js Express Docker Kubernetes Redis Celery Nginx GitHub Actions NVIDIA CUDA

Frontend & Databases

React.js Next.js Three.js TailwindCSS Socket.IO PostgreSQL MongoDB Neo4j Elasticsearch SQLite

POSITIONS OF RESPONSIBILITY

Gen-AI Head

Jan 2025 – Present

Karunya Innovation & Design Studio (KIDS)

AI workshops, LLM research, hackathon mentorship

Coordinator

Jan 2025

K-Hacks Hackathon Club, KITS

Inter-college hackathons and technical events

ACHIEVEMENTS

SIH Shortlisted

PGR-SR: Physics-Guided Residual Super-Resolution for Thermal IR Imagery — RMSE: 0.088K, PSNR: 41.44dB, SSIM top-tier

CV Excellence

100% defect detection rate (40/40) using Laws' Texture Filters; 9-class plant pest detector with MobileNetV2

Open Source Contributor

SurfSense & Scoratis: active community, Discord channels, external contributors; 30+ public AI repositories

CERTIFICATIONS

Vector Databases: Embeddings to Applications

DeepLearning.AI

LangChain for LLM Application Development

DeepLearning.AI

Tools & Agents with LangChain

DeepLearning.AI

ChatGPT Prompt Engineering for Developers

DeepLearning.AI

Foundations of Cybersecurity

Google

Programming Essentials in Python (PCAP)

Cisco / PCAP

Deep Learning with PyTorch: GradCAM

Coursera

Life Expectancy Prediction using Machine Learning

Coursera

Programming Essentials in C (CLA)

Cisco / CLA

GET IN TOUCH

Feel free to reach out if you have any questions, collaboration opportunities, or just want to connect!

Phone: +91-8891201402