Piecewise

Complex problems, manageable pieces

Mathematics · Data Science · Complex Systems

Founding ML Engineer at Mendria, building AI underwriting infrastructure that prices hidden climate risks on physical assets. Previously scaled data science at SmartHelio (YC-22). I work at the intersection of research and production—uncertainty quantification, risk modeling, and ML systems for real-world decision-making.

Journey

Academic foundation to professional growth in data science

BA (Honors) Mathematics

Dr. B.R. Ambedkar University Delhi (AUD) New Delhi 2015–2020
  • Interdisciplinary liberal arts curriculum combining mathematical rigor with social sciences and humanities.
  • Semester-wise merit scholarships for academic distinction (Top 5 rankers).
  • Built foundations in analysis, algebra, and quantitative thinking with attention to context and critique.

MSc Mathematics with Data Science

Institute of Mathematics and Applications Bhubaneswar, Odisha 2020–2022
  • Awarded NBHM scholarship for academic excellence (Top 3 rankers).
  • Founded and led the Data Science Society; organized workshops and interdisciplinary collaboration.
  • Bridged pure mathematics with applied ML and data-driven modeling.

Mendria — Founding ML Engineer

Remote Jan 2026–Present
  • Building AI underwriting infrastructure that detects, quantifies, and prices hidden climate risks on physical assets—insurance drift, energy volatility, regulatory penalties, and asset value decline.
  • Designing ML models for climate risk scoring, loss projections, and uncertainty-aware financial outputs for REITs, lenders, and insurers.
  • Translating domain logic into production-ready modeling pipelines across climate, real estate, and financial risk.
Complex Systems Risk & Decision Modeling Uncertainty Quantification

SmartHelio (YC-22) — Data Science

Noida, India Sep 2022–Nov 2025
3→10
team scaled
3000+
PV assets
delivery velocity
  • Scaled Data Science function from 3 to 10, owning hiring, mentorship, and execution discipline.
  • Architected fault detection, performance modeling, and weather simulation for solar assets.
  • Introduced testing practices, CI/CD, and modular analytics libraries for reliability at scale.
Energy Systems Performance Modeling Anomaly Detection

How I Work

Bridging Research & Engineering

I'm driven by a deep interest in how data shapes knowledge, decisions, and infrastructure. I enjoy working at the intersection of research and engineering, where algorithms meet real-world systems to solve practical problems. I care about engineering rigor too: clear interfaces, clean code, tests, and reproducible pipelines.

Interests

Technology

I'm passionate about understanding and developing intelligent systems at the intersection of society, data, and science.

AI & Machine Learning Systems
Data-Driven Decision Making
Socio-Technical Systems

Academic

Reading across technology, society, philosophy, and mathematics, with a focus on the epistemology of data and the science-society interface.

Philosophy
History
Mathematics
Culture
Complex Systems
Networks

Research

I'm inclined towards studying complex dynamical systems around humans, urban environments, and emerging technologies through networks and data-driven approaches.

Member of the Network Science & Graph Theory Group at IIIT Kottayam, mentored by Dr. Divya Sindhu Lekha. Completed two reading projects in complex networks. People

Human Dynamics
Urban Complex Systems
Cultural Networks
AI Technology Impact

The Common Thread

What connects all these interests is a fascination with emergence — how simple rules and interactions give rise to complex behaviors and patterns. Whether it's in technological systems, social structures, or natural phenomena, I'm drawn to understanding how complexity arises from simplicity and how we can leverage this understanding to build better systems and make more informed decisions.

Featured Projects

Selected Work (Industry)

Fast Diagnostic Service
Cloud-based PV health assessment microservice (Jan 2025)
  • Self-serve uploads (CSV/Excel) → automated diagnostics, KPIs, loss quantification, and benchmarking.
  • Robust input validation + timestamp repair; optimized for large plant datasets.
  • Enabled 300+ new client engagements in ~3 months by replacing multi-hour manual analysis.
Cleaning Optimization Engine
Cost–loss optimization + scenario simulation (2024)
  • Decision pipeline across soiling estimation, rainfall modeling, and PV generation forecasting.
  • Constrained optimization to recommend high-value cleaning dates under operational rules.
  • Deployed across 300+ utility-scale PV assets as a commercial offering.
formatify-py
Datetime inference library for messy timestamps (May 2025 – Present)
  • Pure-Python structural parsing engine for heterogeneous timestamp formats.
  • Validation suite + CI pipeline across thousands of variants.
  • View on GitHub

Networks Lab

Interactive network visualizations and small experiments in graph-based reasoning. Start with a directed artist-relations network.

Explore

Let's Connect

Open for collaborations and interesting discussions

Explore My Resume