AI Data Engineer · Data Systems
I build production systems that process critical data at scale. Applying that to energy markets: ETRM software, commodity analytics, and the Network infrastructure in Telecom.
From my childhood watching my father work in Energy plants and growing up learning and listening his stories of the same, I felt in love with the industry and its fundamental stand in the society.
As and AI Engineer I've spent two years building production infrastructure for telecom operators at national scale: authentication systems, compliance platforms, real-time data pipelines handling millions of events per hour. End-to-end, sole developer, live traffic This helped me connect the dots of the modern day requirement of Digital Infra and Energy Sector.
Now I'm building in energy. Enerra is my open-source ETRM — trade capture, position management, and risk tracking for commodity markets, built on Django and integrated with MCX and IEX. The quant modelling work applies macroeconomic and ESG signals to Indian commodity markets using LLMs.
Trade capture, position management, P&L tracking, and risk systems. Built Enerra, an open-source ETRM on Django with MCX/IEX integration covering futures, hedging, and commodity workflows.
Systems that ingest, process, and act on data at scale. Built for Bharti Airtel at 1M+ QPS across 10+ national sites. Designed for the volume and velocity production environments actually produce.
Quantitative modelling for Indian markets using LLMs, macroeconomic indicators, ESG signals, and price action. Translating raw market data into decisions.
LangChain, LangGraph, RAG pipelines, agentic workflows. Built intelligence layers on top of production data — anomaly detection, forecasting, and natural language interfaces.
DCF, sensitivity analysis, capital budgeting. Infrastructure and project finance economics. Built for energy asset analysis — understanding the capital behind the infrastructure.
Production-grade systems built to satisfy regulators from day one, not retrofitted. DoT/TRAI compliance, audit-ready architecture, designed for operators who can't afford to fail inspection.
For energy startups, IPPs, and renewable operators who generate data they can't act on. I build dashboards, forecasting tools, and analytics pipelines that turn generation, pricing, and grid data into decisions.
For trading firms and energy companies that need custom trade lifecycle tooling or integration work around existing ETRM systems. I understand both the engineering and the trading workflows.
For telecoms, ISPs, and IoT operators who need infrastructure that doesn't break. I've built systems handling 1M+ QPS in production — real traffic, live compliance, sole developer.
An independent review of your data architecture, trading system design, or compliance posture — with clear recommendations from someone who has built it in production.
An open-source Energy Trade and Risk Management platform built on Django and Django REST Framework. Covers trade capture, position management, futures and hedging workflows, and market data integration with MCX and IEX. Built to understand — from the inside — how commodity trading desks manage the full trade lifecycle.
View on GitHub →Research paper implementation adapted for Indian commodity and equity markets. Builds separate logical engines factoring in ESG signals, macroeconomic indicators, and price action using LLMs — producing market views that go beyond simple price-action models.
Telecom operators must submit IP session logs to the Department of Telecommunications on demand — but this startup had no compliant IPLMS or AAA infrastructure, and buying a vendor solution was not financially viable at their scale. I designed and built the entire IPLMS platform and DIAMETER protocol stack from scratch: Gx for policy control, Gy for online charging, and Rx for resource reservation — each interface implemented and tested against live production traffic. Compliance logging, retention windows, and regulatory reporting were designed in from day one, ensuring every DoT/TRAI requirement was satisfied before the first commercial subscriber was onboarded. The system went live handling real operator traffic and remains the authentication and compliance infrastructure the operator runs their network on today.
Facility managers and NOC teams work from siloed systems that generate raw alarms without context — manual fault investigations take 40–60 minutes per incident. This platform unifies BMS sensors (HVAC, energy, access, elevators) and NMS devices (routers, BTS/RAN, servers, UPS) under a single Chat-to-Fix interface: operators describe a fault in plain English, the LLM+RAG engine retrieves device history and vendor runbooks, and returns a specific resolution in under 90 seconds. An LSTM model predicts hardware failures 24–72 hours ahead with 87% precision and auto-generates prioritised work orders; a GNN traces alarm cascades to the single root cause in under 8 seconds.
Telecom operators are entirely reactive — they learn about coverage failures when users complain, after the damage is done. This platform inverts that model: a 4-model ML ensemble (LSTM/TFT for time-series KPI forecasting, GNN for topology cascade propagation, Spatial RF Model for coverage hole prediction, Isolation Forest for real-time anomaly detection) fuses RAN KPIs, IPLMS session flows, weather/terrain data, UE probes, and NMS alarm history to produce a 0–100 outage risk score per cell site every hour. SHAP explainability surfaces the top 3 contributing risk factors; an LLM narrates the risk in plain English and recommends a proactive intervention — before a single user drops a call.
If you're at an energy trading firm, a commodity analytics team, or a renewable operator — and you need someone who understands both the engineering and the markets — I'd like to hear what you're working on.
I'm open to freelance engagements, project-based consulting, and longer-term technical partnerships.
Whether you're building energy market infrastructure, need ETRM development, commodity analytics, or production data systems — I'd like to hear what you're working on.
ankitjha8891@gmail.com