Profile banner
Param Thakkar

Param Thakkar

Computer Engineering student, generative AI researcher and open-source developer

ParamT
Mumbai, India
Joined April 2026
Favicon imagegithub.com
Favicon imagehuggingface.co
Favicon imagescholar.google.com

Network

3.6K connections
💻
VJTI Competitive Coders
🎓
VJTI Student Leaders
🔬
Big Tech AI/ML Researchers
💰
VJTI Finance Interns
💊
PumasAI Pharma
📊
...
🧠
Transformer Lab AI

Summary

Generative-AI and LLM practitioner: As a VJTI Computer Engineering student he has focused on generative AI, LLM workflows and RAG-style systems, presenting at JuliaCon 2025 and authoring the JuliaHealthLLM prototype to apply LLMs to domain-specific (health/OMOP) queries. pretalx+2
Active open-source contributor in scientific computing: He has contributed to the SciML ecosystem, including a refactor of OrdinaryDiffEq.jl to use solver sub-packages (work credited in SciML blog posts), improving startup/load performance for differential equation solvers. sciml+1
Hands-on systems builder and project author: Maintains practical ML/AI projects (PDF-to-data converter, product recommendation system) that integrate LLMs, multi-agent tools, and vector DBs; publishes datasets and models on Hugging Face and code on GitHub. github+2
Early-career researcher with multidisciplinary publications: Co-authored papers on quantum/classical ML for flood prediction and benchmarks for RL-based quantum architecture search, indicating a blend of applied ML, quantum computing, and benchmarking research. arxiv+2

Work

Education

Projects

Writing

BenchRL-QAS: Benchmarking reinforcement learning algorithms for quantum architecture search

July 1, 2025

BenchRL-QAS presents a unified benchmarking framework evaluating multiple RL agents across variational quantum algorithm tasks (2–8 qubits), proposing a composite ranking metric combining accuracy, circuit depth, gate count and training time.

Favicon imagearxiv.org

Personalized recommendation systems using multimodal, autonomous, multi agent systems

October 1, 2024

Work on multimodal autonomous multi-agent recommendation systems (listed on author profiles).

Favicon imagehuggingface.co

Flood Prediction Using Classical and Quantum Machine Learning Models

July 1, 2024

Study comparing classical ML and quantum machine learning approaches for daily flood forecasting on Germany's Wupper River, evaluating training time, accuracy and scalability of hybrid models.

Favicon imagearxiv.org

Hobbies

Exploring world models, reinforcement learning, and GPU-accelerated experiments. nitter+1