Hi, I'm Prayas Jadhav
AI/ML Developer & Full-Stack Engineer building intelligent systems that think, learn, and scale.
About Me
Dynamic Developer with expertise in AI, Python, JavaScript, and ReactJS.
AI/ML Intern
Solar Industries India Ltd.
Dec 2025 – Present
B.Tech in AI
G. H. Raisoni College
I believe in bridging the gap between cutting-edge AI research and production-grade software. Skilled in fine-tuning LLMs, AI model optimization, deployment workflows, and front-end development using modern ReactJS and ES6+ practices.
Prayas Jadhav
AI/ML Developer
Nagpur, India
IST (UTC+5:30)
2+
Years Exp.
5+
Projects
7
Certifications
Skills Universe
Experience
AI/ML Intern
Solar Industries India Ltd.
- ▸ Fine-tuned and deployed in-house LLM for text-generation pipelines
- ▸ Developed advanced RAG system with semantic chunking for knowledge-graph and vector store
- ▸ Building production-grade text-to-SQL agentic system for real-time data analysis
- ▸ Building unified platform with Django backend and React+Vite frontend
AI/ML Intern
ImmverseAI
- ▸ Developed RAG system using graph databases for efficient QnA retrieval
- ▸ Built agent-based pipelines for automated Q&A generation with BERT and LLaMA 3.3
- ▸ Optimized models via hyperparameter tuning and quantization using Hugging Face Transformers
Featured Projects
A showcase of work in Large Language Models, Retrieval-Augmented Generation, and full-stack engineering.
CodeGraphX
A local, token-efficient codebase indexing engine for AI agents. Uses Tree-sitter to parse code incrementally, builds a full dependency graph in a high-density TOON format, and exposes it via an MCP server — eliminating costly file-scanning loops. Features O(1) symbol lookups via Bloom filters, real-time watch mode, an interactive dashboard, and session diff tracking.
O(1)
Symbol Lookup
TOON
Token Format
6
MCP Tools
- Eliminates expensive file-scanning: agents get instant architectural context
- Bloom filter index for sub-millisecond symbol existence checks
- MCP server integrates with Gemini CLI, Claude Desktop, Cursor & Windsurf
- Session diff & branch comparison for change impact analysis
- 100% local execution — no telemetry, no cloud sync
Advanced Hybrid-RAG
Production-grade RAG combining vector search with Neo4j knowledge graph for complex document layouts. Handles hybrid retrieval, semantic chunking, and multi-hop reasoning at scale.
GPT from Scratch
Ground-up implementation of GPT-2 with BPE tokenization, multi-head attention, RoPE, and RMS normalization.
AI Movie Recommender
Conversational movie recommendation system fine-tuned using LoRA on stablelm-2-1-6b with a custom Bollywood dataset.
Grammar Voice Classifier
Lightweight DistilBERT model fine-tuned to classify sentences as active or passive voice with LIME interpretability.
Movie Search App
React-based movie search built with Vite. Integrates TMDB API with Appwrite for tracking popular searches.
Let's Work Together
Have a project in mind or just want to say hi? Feel free to reach out! I'm always open to discussing new AI/ML or Full-Stack opportunities.