Government Engineering College
8.12 CGPA
Hi, I'm Siddharth — a Deep Learning and Generative AI Engineer focused on building production-grade, real-time voice agents, RAG pipelines, and intelligent multimodal systems. I blend my expertise across ML, systems programming, and deployment to develop solutions that push the boundaries of what's possible with AI.
Education
Work Experience
Upstair Technologies LLP
AI ML Developer
- Chatbot using LLM, LangChain, NVIDIA Models and PostGres for Database.
- Using LangChain Memory, Redis for Conversational Memory.
- Voice Agent Wrapper using RealtimeTTS and gTTS.
- Created the Two Project : "ChatBot for Diamonds" & "Voice Agent Capable System"
Otobit PVT LTD
Quality Analyst / Quality Control
- Day to Day Responsibilities include creating a bugsheet and communicating it with developers.
- Using Chrome Developer tools, Postman API, Flask App and regression testing to push features or software on time.
- Creating product flow and conducting regression testing.
Projects
CKD Prediction
- This project implements a machine learning-based prediction model for Chronic Kidney Disease (CKD) using patient health data.
- It helps in early detection and diagnosis by analyzing various medical parameters.
- Built using Python, with scikit-learn, pandas, NumPy, and Matplotlib for data processing, model training, and visualization.
- The model leverages classification algorithms like Decision Trees, Random Forest, and SVM to predict CKD status with high accuracy.
- Integrated Docker for containerization and DVC for data version control.
Chatbot
- Built a full-stack AI-powered chatbot using FastAPI, PostgreSQL, and LangChain with an NVIDIA/Meta LLM backend.
- Designed a modern, responsive frontend for natural language diamond search and recommendations.
- Implemented robust normalization, async database queries, and multi-turn chat with session support.
- Achieved an 85%+ user query resolution rate, delivering accurate results and summaries for diamond buyers.
- Added comprehensive logging, error handling, and concurrency/load testing for production reliability.
End to End Voice to Voice System for Voice Agents
- Developed a real-time conversational AI voice agent for diamond recommendations using FastAPI, WebSockets, and Redis.
- The system leverages Whisper for live speech-to-text (STT), LangChain+OpenAI for natural language understanding and dialogue management, and gTTS for text-to-speech (TTS) responses.
- Designed and implemented a multi-turn dialogue flow with slot-filling, live transcription, and dynamic TTS playback, enabling seamless, interactive voice-based user experiences.
- Tech Stack: Python, FastAPI, WebSockets, Redis, Whisper, LangChain, OpenAI, gTTS, NumPy
Skills
- Python programming (advanced)
- Data analysis and machine learning (Jupyter, Pandas, scikit-learn, etc.)
- Deep learning frameworks (PyTorch, TensorFlow)
- Natural language processing (NLP) and large language models (LLMs)
- Web frameworks (FastAPI, Flask)
- Database management (PostgreSQL, SQLite)
- Web development basics (HTML, CSS)
- Containerization (Docker)
- Conversational AI, LLMs, voice agent frameworks
- Collaborative coding and version control (Git, GitHub)
- Interactive notebook environments (Jupyter)
- Potential exposure to healthcare AI/data science