Hey, I’m Pranav.
I build AI/ML systems and the software around them. RAG pipelines, forecasting models, production frontends. CS graduate and published IEEE researcher.
Experience
- Built responsive, user-facing frontends in HTML, CSS, and JavaScript for production applications.
- Developed a stock price forecasting model in Python using time-series methods to predict market movements.
- Engineered a hybrid CNN–BiLSTM–Attention forecasting model in Python (TensorFlow, Keras) that cut RMSE ~25% (12.40 vs 16.49) and raised R² to 0.87 (from 0.63) against the strongest baselines on real-world electricity-price data.
- Built end-to-end ML data pipelines fusing quantum-chemistry, epidemiological, and genomic datasets into unified spatiotemporal models using Python, Pandas, NumPy, and Scikit-learn.
- Published a peer-reviewed paper in IEEE Data Descriptions (2025), curating a standardized dataset spanning 40 markets across 5 continents.
- Built a RAG-based AI health chatbot end-to-end with a team (frontend and backend) using LangChain, OpenAI APIs, and a vector database to answer user symptom questions with retrieval-grounded responses.
- Applied prompt engineering, embeddings, and vector search to improve answer relevance and reduce hallucinated responses in the end-to-end system.
- Developed and tested frontend interfaces for a HIPAA-compliant medical application alongside the engineering team.
- Built and automated data-processing pipelines over large datasets using Python, Pandas, and Dataiku, applying data-warehousing and ETL techniques for downstream analytics.
- Wrote SQL queries and cleaned, transformed, and validated large datasets to prepare reliable inputs for analysis.
- Shipped dashboards that translated analytics into decisions for non-technical stakeholders across teams.
Projects
Commutr
2025 – PresentFounded a student commuting platform connecting off campus students with student drivers through schedule based rides, leading product design, user research, and development. Ran surveys on commuting habits, designed the end to end product in Figma from wireframes to interactive prototypes, and scoped a 150–300 rider pilot with subscription based revenue projections.
VALS ↗
2026Led a team of 4 to build a realtime VR label autocomplete system with a Trie based prefix search supporting full word and abbreviation matching, loading anatomy data dynamically from CSV.
Maargdarshak ↗
2024A guidance tool with a Streamlit interface tuned for usability and accessibility, wiring Python logic to the OpenAI API for realtime interactivity.
Arctic Mercury Modeling ↗
2025An ML assisted framework modeling mercury reduction, reemission, and diffusion across Arctic snow and ice, blending quantum chemistry and molecular dynamics with data driven models.
Restaurant Management ↗
2023A dynamic order management platform with a staff interface for categories, orders, and customer records, backed by MySQL for reliable storage and retrieval.
Research
UniElecPrice: A Unified Cross Regional Day Ahead Electricity Price Dataset
Introduced a globally unified dataset of day ahead electricity prices drawn from 40 markets across 5 continents, enabling standardized forecasting research. Developed and validated a hybrid CNN/BiLSTM/Attention model that outperformed LSTM and Transformer baselines.
Toolkit
Let’s build something.
The fastest way to reach me is email. Always happy to talk shop.
p.pranavbalachander@gmail.com