I am an AI/ML & Autonomous Systems Engineer specializing in production Multi-Agent architectures, Vision-Language Models (VLMs), and enterprise RAG pipelines. I focus on designing intelligent systems that bridge real-world business ambiguity with deterministic reliability through multi-phase reasoning loops (Plan → Execute → Reflect → Compile).
- 🤖 Autonomous Multi-Agent Systems: Designing agent swarms, dynamic TODO task planners, self-reflection quality gates, and circuit-breaker safety loops.
- 👁️ Computer Vision & Multimodal AI: Building bidirectional Text-to-Image diffusion studios, real-time sign language translators, and VLM captioning engines (ViT + GPT-2, YOLOv8).
- ⚡ Enterprise Generative AI: Engineering low-latency LLM inference pipelines with Google Gemini, Groq, and Llama 3 backed by vector search and relational schema awareness.
- 🚀 Full-Stack AI Engineering: Production-ready backend APIs with FastAPI, Flask, Docker, and PostgreSQL, accompanied by responsive modern interfaces.
📍 Location: Nagpur, Maharashtra, India
💼 Portfolio: harshrameshnerkar.vercel.app
📫 Direct Contact: harshrameshnerkar@gmail.com • +91 7522946205
| Project | Description | Key Tech Stack |
|---|---|---|
| 🤖 Python AI Engineer – Autonomous Agents | Enterprise autonomous document generation agent. Decomposes ambiguous business prompts into dynamic 6-stage TODO plans, performs automated gap audits, and compiles publication-grade Word deliverables. | Python, FastAPI, Gemini API, Groq, Pydantic v2, python-docx, Tailwind CSS |
| 🖼️ CaptionAI Multimodal Studio | Bidirectional AI vision platform featuring Text-to-Image diffusion across 8 art styles, Vision-to-Text captioning with TTS audio narration, multimodal concept remixing, and model training hub. | PyTorch, ViT, GPT-2, HuggingFace, AI Diffusion, gTTS, Flask |
| 🛡️ Agentic AI Systems | 20-Day production engineering suite for autonomous SRE agents, multi-agent swarms, self-healing workflows, hybrid RAG, and HITL safety boundaries. | Python, asyncio, Llama 3, Hybrid RAG, Circuit Breakers |
| 🤟 ISL Translator | Real-time Indian Sign Language (ISL) and hand gesture recognition translator with bilingual text-to-speech audio and interactive web dashboard. | PyTorch CNN, OpenCV, MediaPipe, Bilingual TTS, Web Studio |
| ⚡ Universal Text-to-SQL AI | Multi-table relational schema-aware SQL generator with automated self-correction, execution sandbox, and multi-dialect SQL support. | Python, Google Gemini, Groq, SQLite, PostgreSQL |
| 🌿 MindWell Companion | AI-powered mental wellbeing companion with multi-agent orchestration (Coach, Safety, Memory), crisis detection, and conditional RAG. | Llama 3, Neon PostgreSQL, RAG, Streamlit |
(Sep 2026 – Present)
- Architecting enterprise-grade Autonomous Multi-Agent Systems with multi-phase lifecycles (dynamic planning, self-reflection, and quality gating).
- Engineering resilient RAG pipelines and structured reasoning systems using modern LLMs (Gemini, Llama, Groq) with circuit breakers and HITL safety.
- Developing automated business document generators and agentic AI tools adopted for production workflows.
(Jan 2026 – Apr 2026)
- Architected a multi-agent AI platform with distinct agent personas (Orchestrator, Wellness Coach, Safety Gatekeeper, and Memory Manager).
- Implemented a hallucination-resistant RAG pipeline with Neon PostgreSQL and pgvector for grounded mental wellbeing conversations.
- Conducted prompt optimization and token efficiency benchmarking across open-source and proprietary LLM backends.
(May 2025 – Jul 2025)
- Built a real-time Indian Sign Language Translator using CNN, RNN/LSTM, and OpenCV for bidirectional accessibility.
- Trained and deployed a YOLOv8 Safety Helmet Detection computer vision pipeline for real-time construction site surveillance.
- Created gesture-controlled virtual keyboard interfaces and automated notification bots.
- B.Tech in Computer Science and Engineering
SB Jain Institute of Technology, Management & Research, Nagpur
Cumulative GPA: 8.23 / 10.0
| Domain | Technologies & Frameworks |
|---|---|
| Core Languages | Python (Advanced) • SQL • C++ • JavaScript • HTML5 / CSS3 |
| Autonomous Agents & LLMs | Multi-Agent Architectures • RAG Pipelines • Plan-Execute-Reflect Loops • Google Gemini API • Groq • Llama 3 • LangChain |
| Deep Learning & Vision | PyTorch • TensorFlow • HuggingFace Transformers • Vision Transformers (ViT) • OpenCV • MediaPipe • YOLOv8 • Diffusers |
| Backend & APIs | FastAPI • Flask • Pydantic v2 • python-docx • RESTful APIs • Uvicorn |
| Databases & DevOps | PostgreSQL • Neon DB • SQLite • Docker • Git / GitHub • Linux |
⚡ "Architecting intelligent, resilient, and production-ready AI systems." ⚡
