Headline

AI/ML Practitioner | Blending Critical Thinking with Machine Learning & Data Science • Python • NLP • Computer Vision • Deep Learning • Generative AI

About

From analyzing complex political systems to architecting AI agents across multiple modalities. My path into technology wasn't linear, and that’s my greatest strength. With a background in Political Science and Economics paired with a deep dive into Computer Applications and AI, I approach engineering with a dual lens: building systems that are technically rigorous and contextually aware. I specialize in the Agentic AI stack - utilizing LangGraph, FastAPI, and PyTorch to build autonomous systems that don't just process data but "think" through complex states. I am particularly obsessed with Local-First AI, developing privacy-centric RAG engines and voice assistants that prioritize data security and low latency over cloud convenience. Technical Stack: • Core: Python, Rust, PyTorch, scikit-learn, SQL • AI Architecture: LangGraph, LangChain, RAG (Hierarchical/Dual), Neuro-symbolic AI • DevOps & Tools: Arch Linux, Neovim, Tmux, FastAPI, PostgreSQL, FAISS, ONNX Runtime When I’m not in Neovim, I’m likely exploring the geopolitical significance of international naval bases, diving into game theory, or troubleshooting my Arch Linux dotfiles. I believe the best engineering happens at the intersection of high-stakes strategy and systems-level mastery. 👇 Check out my featured projects below for a deep dive into my recent work with LangGraph, Edge AI, and Neuro-symbolic systems.

Location

Jaipur

Work experience

Artificial Intelligence Engineer at Independent Research & Development, Jun 2024 - Present · 2 yrs 3 mos, • Focused on designing production-ready AI systems with low-latency inference, agentic workflows, and robust state management. • Autonomous Agent Architecture (Nexus Bot): Engineered a LangGraph-based multi-agent system with PostgreSQL-backed persistent memory for long-term contextual recall. • Neuro-Symbolic RAG System (MedPal AI): Developed a clinical decision support system using Dual-RAG and specialized BERT models to handle domain-specific queries while reducing hallucination risk. • Real-Time Conversational AI: Built an emotion-aware voice assistant with real-time intent recognition, Silero VAD, and Kokoro TTS for low-latency speech interaction.

System Engineer at Independent Research & Development, Jun 2024 - Present · 2 yrs 3 mos, Self-Directed / Open Source •Engineered nexus-bot, a LangGraph + MCP + PostgreSQL stateful autonomous agent with persistent tool registry and multi-turn episodic memory. •Built MedPal AI, a neuro-symbolic Dual-RAG clinical decision support system routing queries through dense FAISS retrieval and a logic-gated medical execution path — 40% hallucination reduction on domain-specific queries. •Developed a fully local emotion-aware voice assistant achieving sub-100ms barge-in latency on a dual-core i5 via Rust VAD processor (WASM/​ONNX), Faster-Whisper STT, and Kokoro-ONNX TTS — zero cloud dependency. •Architected CodexEngine v3, a multi-LLM agentic RAG with LangSmith distributed tracing and RAGAS evaluation harness for cross-version golden dataset regression testing.

Technical Support Executive at TP, Jul 2023 - Apr 2024 · 10 mos, Delivered email and chat-based support for billing, invoicing, and technical issues. Resolved complex technical problems while maintaining SLA and KPI benchmarks. Supported Japanese clients with software and network troubleshooting. Managed tickets using Salesforce and Zendesk.

Education

Minor in Data Science and Artificial Intelligence, Artificial Intelligence, Indian Institute of Technology, Mandi

Bachelor's in Computer Applications, Computer Programming, Specific Applications, Jaipur National University

Public links

github: https://github.com/anmolsharma152