Zero-Shot Cell Annotation
Built zero-shot classification pipelines with foundation models (such as Geneformer and LangCell) for healthcare datasets.
About
Healthcare AI · Clinical NLP · Foundation Models · Agentic & Efficient ML
I build trustworthy AI for healthcare and science: conversational agents, clinical NLP, evaluation & safety, and foundation models — with interests in efficient ML and large‑scale data.
What I Do
- Build AI-driven systems for healthcare products and research
- Work on clinical NLP, AI agents, evaluation & safety, and foundation models
- Research in mental health; selected work in computational genomics
Tech Stack
Languages
Python, SQL, C/C++, Java, MATLAB
ML & Data
PyTorch, TensorFlow, scikit-learn, NumPy, Pandas, Scanpy, scvi-tools, UMAP
Web & Tools
FastAPI, Docker, PostgreSQL, Node.js, React, LangGraph, LangChain, LangSmith, CUDA
Evaluation & Safety
Robustness, fairness, reproducibility; evaluation pipelines and safety tooling
Featured Projects
Experience
UCL Centre for Artificial Intelligence – Agentic AI Researcher
October 2025 - Present (in collaboration with Freedom from Torture)
- Prototyping multilingual mental‑health conversational agents using LLMs and agentic AI frameworks (LangGraph, CrewAI).
- Partnering with Freedom from Torture to ensure trauma‑informed, culturally sensitive design and deployment.
- Collaborating with clinical researchers to embed evidence‑based therapeutic strategies into AI systems.
- Developing synthetic clinician‑patient dialogue datasets for training and evaluation.
- Architecting safe, interpretable agentic workflows for reasoning and planning in sensitive contexts.
- Integrating and testing LLM services (ChatGPT, Claude, Llama, Groq, AWS Bedrock) within cloud pipelines.
Secrier Lab, UCL Genetics Institute – AI & Genomics Researcher
March - September 2025
- Built scalable ML pipelines (PyTorch, scvi-tools, Scanpy) analysing millions of single‑cell RNA‑seq datapoints across cancer types.
- Benchmarked foundation models (scBERT, scGPT, Geneformer, LangCell) for classification and state prediction with strong Accuracy, F1, AUROC, AUPRC.
- Developed zero‑shot annotation techniques and improved cross‑dataset generalisation under batch effects using UMAP.
- Produced actionable insights with a reproducible framework; pre‑print planned for 2026.
Open-Source Assistive Devices (OPAD) - President
University of Nottingham, 2022 - 2024
- Directed student-led projects building assistive and medical devices across hardware, software, and analytics.
- Applied ML to improve device performance and automate data collection aligned to NHS needs.
- Secured funding and coordinated cross-disciplinary engineering teams.
40seconds.org - Founder and CEO
Non-profit, 2020 - Present
- Founded a global non-profit promoting mental health awareness and human rights for youth.
- Led initiatives including articles, research, and volunteer programs across 19 countries.
- Raised funds for Covid-19 relief and sustained operations with a student-led team.
Honours
- Vice-Chancellor’s Medal (Top 0.1%), University of Nottingham – Highest award for exceptional contributions as author, philanthropist, and student leader (2024)
- EEE Undergraduate Studies Award, University of Nottingham – Academic and departmental contributions (2024)
- Nottingham Advantage Awards, University of Nottingham – Student Leader (2024); OPAD & Peer Mentor (2023)
- Union Prize & Student Representative of the Year, University of Nottingham (2022)
- Top 6 Teen Heroes, India – National recognition for social impact and leadership (2021)
- Speaker invitations – World Summit AI, TEDx, and university conferences on AI and healthcare innovation
CV & Highlights
Education
- University College London (UCL) – MSc Artificial Intelligence for Biomedicine and Healthcare (2024 - Present)
- The University of Nottingham – BEng (Hons) Electronic and Computer Engineering (2021 - 2024)
Highlights
- Healthcare AI: clinical NLP, conversational agents, evaluation & safety
- Foundation models and agentic/efficient ML for large‑scale data
- Experience: PyTorch, TensorFlow, Scanpy, scvi-tools, FastAPI, React, LangGraph/LangChain
- Open-source contributor; reproducible workflows and benchmarking