I am a Data Scientist Manager and AI Engineer passionate about building real-world AI-powered solutions. My expertise spans Generative AI, Agentic Systems, Computer Vision, Industrial IoT, and Robotics, where I combine research-backed approaches with practical, production-grade deployment.

I have designed and delivered projects ranging from autonomous AI firmware for microcontrollers and multilingual conversational agents to real-time analytics dashboards and AI-driven mobile applications. My focus is always on building scalable systems that bridge the gap between software intelligence and physical hardware to create real impact.

I actively work with:

• Agentic AI & LLMs: CrewAI, LangGraph, Google Gemini, Groq, Llama 3
• Edge & Hardware: ESP32, FreeRTOS, PlatformIO, TFLite
• Cloud & MLOps: AWS, GCP, MLFlow, Docker
• Full-Stack AI: FastAPI, Flask, Streamlit, Flutter

My strength lies in end-to-end AI development, from building robust ML pipelines to developing user-centric interfaces and embedded hardware solutions, always driven by continuous learning and a commitment to excellence.

📧 Let's Connect: abhimanyus1997@gmail.com

• MicroClaw - Agentic Firmware for ESP32 (2026)
  Tech Stack: C++, Python, FreeRTOS, Gemini/Groq API, FastAPI
  Developed agentic firmware transforming an ESP32 into an autonomous physical AI node. Engineered a ReAct-loop tool-use system utilizing FreeRTOS for dynamic GPIO scripting, WiFi/BLE scanning, and persistent LittleFS memory. Built an ecosystem featuring an on-device web chat and a FastAPI desktop manager for OTA updates and monitoring.
  🔗 View Live Docs

• HealthBuddy - AI Wellness Companion (2026)
  Tech Stack: Flutter, Groq API (Llama 3), Google Health Connect
  Developed a privacy-first mobile health app that integrates Google Health Connect to sync biometrics. Built NutriGPT, a conversational AI providing low-latency wellness advice and automated, personalized diet chart generation using a secure "Bring Your Own Key" (BYOK) architecture.

• Resume JD Matcher (2025)
  Tech Stack: Python, LangChain, Gemini 2.5 Flash, Streamlit, Plotly
  Engineered an AI recruitment tool utilizing Gemini 2.5 Flash for skill extraction and TF-IDF scoring. Automated PDF parsing and generated dynamic radar charts via a Streamlit dashboard for precise candidate-job alignment and custom PDF report generation.

• LangChain One (2023)
  Tech Stack: LlamaIndex, LangChain, EasyOCR, YOLOv8
  Created a multi-modal chatbot using LlamaIndex and EasyOCR for image-based data extraction, alongside hybrid RAG pipelines for vision and text inputs. Utilized YOLOv8 for tabular data detection with Few-Shot Prompting for enhanced data accuracy.
  🔗 View Repository

• Article: "The Evolution of Object Detection, YOLO & NAS Revolutionizing Computer Vision"
  Published on IndiaAI.gov.in
• Paper Presentation: "Subaqueous Crack Detection using Fusion-Based Image Enhancement for Dam Structure"
  Presented at the 4th International Conference on River Corridor Research and Management (ICCRM) 2024, hosted by IIT Guwahati & IIT Jammu

Let's connect to build innovative solutions together!