Pragnyan Ramtha

about me.

I design maintainable, production‑grade AI systems and can comfortably work with deep cloud infrastructure when needed. I learn new tools fast and use AI as a force‑multiplier in my coding, desigining, and research loops, which lets me move much faster while keeping systems reliable.

experience.

• AI Engineering Intern Remote

  at, Reputation-DAO

  Aug 2025 – Jan 2025

  • Architected a GCP serverless backend achieving 99.9% uptime by leveraging Cloud Functions and Cloud Run for production-grade AI orchestration.
  • Reduced inference latency by 50% across support workflows by engineering a Gemini API response system with optimized prompt caching.
  • Boosted accuracy and trust by developing a RAG pipeline utilizing semantic search for real-time documentation retrieval and source attribution.

  • GCP
  • Gemini API
  • RAG
  • Serverless
  • Python

• Open Source Developer Remote

  at, Various Open Source Projects

  2024 - Present

  • Actively Contributed to 25+ open-source projects, across various organizations.
  • Refactored and enhanced the codebase to boost maintainability, achieving an 80% developer satisfaction rate.
  • Winner of IEEE Summer of Code (IEEESoC) Hackathon 2025, for my open source contribution to multiple projects.

  • Python
  • TypeScript
  • Git
  • Docker
  • CI/CD

projects.

• AIMO-3: Efficient Reasoning via LLM Fine-Tuning

  • Fine-tuned Phi-4 (14B) on CoT and TiR datasets to optimize multi-step problem solving and tool-use efficiency.
  • Achieved 90% accuracy on reasoning benchmarks, rivaling 70B parameter models while utilizing significantly fewer compute resources.

  • Phi-4
  • Fine-tuning
  • CoT/TiR
  • PEFT

• Personality Clone

  • Fine-tuned a Large Language model, leveraging PEFT (QLoRA) and contrastive learning on private conversational data to emulate personal response style.
  • Implemented a siamese network architecture with cosine similarity loss, which improved semantic embeddings and achieved 92% accuracy in replicating my response style, a 28% improvement over baseline models.

  • TensorFlow
  • Python
  • CUDA
  • Transformers

• Autopilot

  • Engineered an AI-driven OS automation system leveraging function calling and tool-use paradigms to execute complex natural language tasks to achieve low-level automation.
  • Built a Reasoning + Acting agent framework with command sandboxing, reducing execution errors and achieving 45% faster task completion than manual workflows.

  • Python
  • LLM Agents
  • APIs

technical blogs.

• Training a lightweight personality clone August 22, 2025

  Read “Training a lightweight personality clone” 6 min read

  Dataset prep from chats, contrastive fine-tuning, and evaluation pitfalls.

• How I’m building Autopilot September 5, 2025

  Read “How I’m building Autopilot” 5 min read

  Notes on architecture choices, safe command execution, and early UX experiments.

• CheatSheet AI – ranking experiments September 1, 2025

  Read “CheatSheet AI – ranking experiments” 4 min read

  Trying hybrid retrieval + LLM reranking for better meta-search relevance.

technical skills.

• Languages:

  Python, TypeScript, Bash, C

• Certifications:

  Machine learning certification (Stanford), CS50: comp. Sci. (Harvard University)

• AI/ML:

  PyTorch, Transformers, Unsloth, Scikit-learn, PEFT/QLoRA, RAG

• Infra:

  GCP, Azure, Docker, Linux (Arch), Git

• Achievements:

  • Winner, IEEE Summer of Code (IEEESOC) Hackathon 2025
  • Winner, Empathy Encryption Hackathon 2025
  • Winner, Daydream Hyderabad @ Hackclub 2025
  • Top 0.5% Finalist, Shell AI Hackathon 2025

• Tool kit:

  uv, Neovim, Arch Linux