Just a Human In The Loop

• Building autonomous agents so they can take over the boring tasks (and maybe the world someday)
• Crafting AI workflows that pretend to understand what I actually want
• Fine-tuning LLMs — because the base model clearly didn't get my sarcasm
• Using MCP and RAG to confuse both machines and humans alike
• Working on an autonomous stock trading environment — teaching AI to lose money faster than I can

• Generative AI Frameworks: LangChain, Llama-Index, HuggingFace — teaching machines to talk so they can ignore me better
• Agentic AI Frameworks: LangGraph, OpenAI Agents SDK, CrewAI, AutoGen, MCP — giving AI just enough freedom to cause chaos
• LLM Engineering: RAG, Fine-Tuning, Context Engineering — because apparently, telling AI what I want is rocket science
• LLMOps / AIOps — “trust me, it scales”
• Machine Learning & Deep Learning Algorithms — the wizardry behind all my AI "accidents"
• Natural Language Processing — making computers pretend they get human nonsense
• Data Structures & Algorithms in C++ — because sometimes you need to make things needlessly complicated
• Almost all core Engineering subjects — a lifetime subscription to "I survived engineering school" therapy

• MAITRI AI – Intelligent Psychological Companion (Smart India Hackathon): Led a 5-member AI crew building a space-ready psychological companion that reads astronauts’ voices and faces in real-time, detects emotions with 92% accuracy, and responds naturally — basically an AI therapist keeping space crews sane while boosting mission efficiency by 35%.
• BioSage – AI Knowledge Engine (NASA Space Apps 2025): Built a multi-agent AI knowledge engine that ingested 600+ space-biology papers and all NASA official domains, delivered 98% citation accuracy, and let researchers pull insights 80% faster
• Built a RAG-based chatbot deployed on my portfolio site to answer recruiters' and stalkers' questions — and promptly snitch if they get weird
• Created a Multi-Agent Deep Research System capable of parallel information retrieval, reflection-based reasoning, and automated generation of structured research reports with improved coherence and depth. — because one AI just wasn't enough chaos
• Assembled a multi-agent AI engineering team with a lead, frontend, backend, and tester — all working tirelessly to do my bidding
• Fine-Tuned Llama-3.1-8B for predicting Product prices by their descriptions, with Mean Error of $47, outperforming traditional ML techniques for the same.
• Built a Self-Organizing AI Idea Refinement Ecosystem - an army of AI agents arguing with each other until they ‘accidentally’ invent something smarter than me. Zero human intervention, maximum existential dread.
• Developed an AI agent that spawns other AI agents who argue and collaborate to optimize ideas — basically my own little AI soap opera
• Built an autonomous SDR that writes and sends cold emails — so I can ghost people professionally
• Completed a hell lot of toy ML and deep learning projects — but don't let that fool you, I've got concepts down cold

• I want to do something for social good — because saving the world looks better on LinkedIn
• Work with experienced pros in AI and Data Science — so I can pretend I know what I'm doing
• And eventually, create commercial projects — because apparently, AI skills pay the bills too