Available for internships & freelance

Building Products,
Not Just Projects.

AI evaluation, LLM fine-tuning, and full-stack development. I ship real-world products with React, Python, FastAPI, and Supabase.

India BTech CSE, 2026
About

Self-driven builder, learning by shipping.

I'm Ashutosh Samanta — a builder focused on creating real-world digital products. I work across full-stack development, AI evaluation, and LLM fine-tuning to turn ideas into working software.

I started building products before entering college, learning by doing rather than waiting for permission. Recently, I fine-tuned Llama 3 8B with LoRA on compliance datasets to build Specter-1.0, an AI auditor for risk and regulatory analysis.

My goal is to build impactful products, contribute to innovative teams, and continuously level up across software engineering, AI evaluation, and product design — where craft meets velocity.

Focus
AI Evaluation + Full Stack
Approach
Ship → Evaluate → Improve
Status
1st-yr BTech CSE
Based in
India
Skills

Core competencies for AI-native products.

From LLM evaluation and fine-tuning to full-stack apps and deployment — here's what I build with.

AI Evaluation

  • AI Response Evaluation90%
  • Prompt Engineering88%
  • Hallucination Detection86%
  • LLM Testing85%
  • Human Feedback Evaluation84%
  • Critical Reasoning90%
  • AI Quality Assurance87%
  • AI Auditing83%

AI & ML

  • Large Language Models88%
  • LoRA Fine-tuning85%
  • Multi-Agent Systems78%
  • Retrieval-Augmented Generation82%
  • AI Memory Systems76%

Development

  • Python88%
  • JavaScript85%
  • TypeScript80%
  • FastAPI78%
  • React82%
  • Next.js75%
  • Node.js72%
  • Express70%
  • Supabase80%
  • SQL70%
  • Git85%

Database & Cloud

  • Supabase80%
  • SQL70%
  • Cloud Storage72%

Languages & Tools

  • Python88%
  • Java65%
  • Git & GitHub85%
  • AI Dev Tools90%
Projects

Real products, shipped.

A few things I've built — designed, coded, and deployed end-to-end.

Specter-1.0

Compliance & Risk AI Auditor.

Fine-tuned Meta Llama 3 8B using LoRA on SEBI, RBI, and CGST compliance datasets to build a regulatory AI assistant. Evaluated outputs for factual consistency, reasoning quality, and domain accuracy throughout the pipeline; iterated on prompts and response design; deployed on Hugging Face for testing and validation.

Llama 3 8BPythonUnslothHugging FaceLoRAFastAPI

Hubbsie

Privacy-first cloud drive platform.

Cloud-drive MVP with authentication, nested folders, uploads, search, sorting, grid/list views, public share links, and file previews — backed by Supabase storage.

ReactNode.jsSupabase

Baard

Secure communication and storage.

A privacy-focused platform concept: secure messaging, file sharing, temporary secure links, offline-first design, and future-ready collaboration features.

ReactNode.jsSupabase
Journey

From first commit to what's next.

Achievements
IITM Road Safety Hackathon 2026
Participant
Shipped real-world apps
Built and deployed full-stack MVPs
Self-driven learner
Continuously improving SWE craft
Timeline
  1. 2025
    Started building projects
    Began crafting web apps and exploring product ideas.
  2. 2026
    Built Hubbsie MVP
    Shipped a privacy-first cloud-drive product end-to-end.
  3. 2026
    Started BTech CSE
    Joined first-year computer science engineering.
  4. Future
    Software Engineer & Product Builder
    Building products at ambitious teams worldwide.
Currently Learning
Full Stack DevelopmentAdvanced JavaScriptReact EcosystemBackend ArchitectureSoftware Engineering PrinciplesCloud TechnologiesAI-Assisted Development
Let's build something

Got an idea?
Let's make it real.

Open to internships, freelance, and collaborations. The fastest way to reach me is email.