Skip to content
View pragnyanramtha's full-sized avatar
🎆
Open to work
🎆
Open to work

Highlights

  • Pro

Block or report pragnyanramtha

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Content in all repositories owned by your account will be closed.
Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. Markdown is supported. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
pragnyanramtha/README.md

Pragnyan Ramtha

AI Engineer · 50+ merged open source PRs

I build autonomous AI agents and fine-tune LLMs for production. Previously #1 on ARC-AGI.

Currently: Ai Researcher @ Parsewave.ai

Skills: Agents · Fine-tuning / QLoRA · RAG · Next.js / TypeScript · Python / Transformers

Portfolio LinkedIn Hugging Face


Projects

Project Outcome Stack
Agent7 No-code agent platform — 1,000+ connectors, 20,000+ tools, multi-hour autonomous sessions Next.js, TypeScript
Personality Clone Style emulation via siamese + contrastive learning — 92% accuracy (+28% vs baseline) TensorFlow, QLoRA, CUDA
Parameter Golf 1.1271 BPB on a 16MB model (XSA4 + EMA + GPTQ-Int6) Python, Transformers
Autopilot AI-driven OS automation with ReAct agents + command sandboxing Python, LLM Agents

Open Source

35 merged PRs across AI/ML frameworks — correctness, security, and production hardening. Full list: github.com/pragnyanramtha

PR Impact
openai-node · #1884 TypeScript SDK correctness
langgraphjs · #2407 Agent-graph / LangGraph JS fixes
pydantic-ai Framework hardening
agno Agent framework correctness
mem0 Memory-layer production patch

Also: observability (opik, openinference, openlit), embeddings (lancedb, pinecone), tooling (promptfoo, fastmcp, dstack, chainlit), plus 6 Dify plugins merged.


Awards

🥇 #1 on ARC-AGI Public Leaderboard — Test-Time Training for fluid intelligence benchmarks

🏆 Winner, IEEE Summer of Code 2025 · Winner, NIAT RAG Challenge · Winner, Empathy Encryption Hackathon · Winner, Daydream Hyderabad

🏅 Top 0.5% Finalist, Shell AI Hackathon 2025 · Finalist, Smallest AI × IIT G


Experience

Founder — Agent7 · 2025–Present
Solo from zero to production. 1,000+ app connectors, 20,000+ tools, autonomous sessions running hours to days.

Software Engineer — Learnable India · Apr 2026–Present
Assistive software for blind learners — screen-reader workflows and AI-powered tools.

Agentic AI Developer — Reputation Dao · Aug 2025–Jan 2026
GCP serverless backend (99.9% uptime), 50% inference latency reduction, RAG pipeline.

ML Engineer (Freelance) — Six Axis Studios · Feb–May 2025
World-model research for architecture workflows, CAD-style design generation.


Education

B.Tech Computer Science (Data Science) — NIAT/MRV University
Machine Learning Specialization (Stanford) · CS50 (Harvard)


Writing


GitHub Streak

Pinned Loading

  1. deepResearch deepResearch Public

    The Deep Research Dify Plugin transforms your AI applications into powerful research assistants by integrating real-time web search capabilities. Using Tavily API, this tool allows your Large Langu…

    Python 5 1

  2. autopilot autopilot Public

    Automate things on your native system with the power of ai.

    Python 3 3

  3. hercule hercule Public

    A browser extension that leverages Ai to scan privacy policies. The system analyzes privacy policies and presents them through an intuitive interface with actionable insights.

    Python 3 2

  4. longmem longmem Public

    A conversational AI agent with persistent, queryable memory that can recall information from turn 1 even at turn 1000.

    Python 2