Market intelligence is fragmented. Every decision is made with partial information.
Rooftop solar, batteries and EVs have turned consumers into millions of small producers. Prices move by the hour, and decisions that used to sit with a handful of grid operators now have to be made by thousands of individual households. Forecasting, bidding and settlement still live in separate tools, spread across separate participants, with no shared view of any given interval.
A forecast in one system, a bid in another, a settlement file in a third. Nobody sees the whole interval, so nobody can act on it in time.
Forecasting, optimisation and execution built as a single reusable platform. It sees the whole interval and acts inside it, autonomously.
Three layers. We build the one in the middle.
Peer-to-peer trading, bulk market execution, flexibility products. Each market plugs into the same core — only the rules layer changes.
Grid, market and weather signals ingested continuously. Every cleared trade is a new data point, so the forecasts keep getting more accurate over time.
One platform, retargeted market by market.
We ship in sequence, not all at once. Peer-to-peer is live. The rest follow the same core.
Prosumers sell surplus solar to neighbours over the DISCOM network. Forecast, offer, match, deliver, settle — every hour, with nobody watching a screen.
Watch the market clear →Thousands of rooftops, batteries and flexible loads aggregated into one dispatchable resource that a grid operator can actually call on.
What we're building →Day-ahead and real-time bidding, portfolio scheduling and imbalance risk — the same optimiser, pointed at bulk power.
What we're building →The infrastructure for decentralised electricity markets is being laid right now.
The participants and the rails are being installed at the same time. That is a narrow window, and it is open. The numbers below are from our first market.
Market sizing: Dataintelo, P2P Energy Trading Market (2026). Scopes differ between studies; figures are research estimates.
A lean team, close to both the models and the decisions they power.
We care about curiosity, integrity, and people who enjoy turning complex systems into something simple. If building cleaner, better-optimised energy systems sounds like your kind of problem, we'd like to talk.
See how we work →Peer-to-Peer
LIVEHouseholds with surplus solar sell it to neighbours who need it, over the existing DISCOM network. The prosumer earns more than the feed-in tariff, the consumer buys green power for less, the DISCOM earns its network charge. Nobody has to manage anything.
Watch an interval clear.
Every hour the platform forecasts each participant's surplus and demand, prices an offer, matches buyers to sellers and books the trade. This is that loop, replayed on representative data.
Jump to the full demo ↓Forecasting and buying, on real and simulated data.
A longer walkthrough: household forecast accuracy against real meters, one buy decision explained, and 84 days of the buying engine running unattended.
Five steps, every interval, all of it on the platform.
Surplus and demand predicted per participant, 24 hours ahead, updated continuously.
The optimiser sets price and quantity for each seller — automatically, within their limits.
Buyers and sellers cleared each interval against the best available price and volume.
Energy delivered over the existing DISCOM network. No new wires, no microgrid required.
Metered, verified and paid, with a record every party can check.
Supply and demand, orchestrated with mathematical precision.
Three models do the work. None of them guess.
Hour-by-hour prediction of what each household will draw over the next 24 hours, built on years of metered consumption and refined every time it is wrong.
Solar irradiance, cloud opacity, panel orientation and soiling — resolved into how much a specific rooftop will actually produce tomorrow, with the uncertainty attached.
The trading layer. Offers, matches and settlement run on the Unified Energy Interface, so trades interoperate with the India Energy Stack rather than sitting in a private silo.
Three parties, no losers.
Run a peer-to-peer pilot in your network.
We work with DISCOMs, installers and aggregators to set up a pilot with a few hundred prosumers on their existing network.
Virtual Power Plant
ROADMAPOperating fleets of distributed assets — rooftops, batteries, EV chargers, flexible industrial load — as a single dispatchable resource a grid operator can call on.
This product is on our roadmap. The forecasting and optimisation core it depends on is live today in peer-to-peer trading; the aggregation, telemetry and dispatch layers described below are still in design.
Thousands of home batteries, charged and discharged together, look like one power plant.
A grid operator can't call on a single household battery when it needs power. It can call on a fleet of thousands, discharging together on command, if it can trust how much they'll actually deliver. That trust comes from forecasting each battery's available charge and predicting what the fleet can deliver as a whole — the same forecasting and optimisation problem our platform already solves for individual prosumers in peer-to-peer trading, applied to a fleet of batteries instead of a household's solar.
Onboard distributed assets, read their telemetry, and build a probabilistic picture of what the fleet can deliver in any given interval.
Turn that distribution into a firm offer the operator can rely on, with the risk of under-delivery priced into it up front.
Allocate the call across assets — cheapest, least disruptive first — then verify delivery and settle each owner's share.
Everything below the market-rules layer. That is the whole argument for building peer-to-peer first.
Wholesale Markets
ROADMAPSame platform, different market rules. Bidding generation portfolios into day-ahead and real-time markets, scheduling storage against price spreads, and managing imbalance risk across intervals.
This product is on our roadmap. The forecasting and optimisation core is live in peer-to-peer; the exchange connectivity and bidding strategies described here are still in development.
A wholesale bid and a neighbourhood offer are the same problem at different scales.
Forecast a quantity under uncertainty. Price it against rules and constraints. Submit before the gate closes. Settle against what actually flowed. Only the rules layer changes between markets — and the intelligence underneath compounds across both.
Day-ahead and real-time bids across a generation portfolio, optimised jointly rather than asset by asset.
Charge and discharge decisions against forecast price spreads, degradation cost and state-of-charge limits.
Deviation exposure priced into the bid itself, so the penalty is a managed cost rather than a surprise.
A continuous decision loop. Every cycle makes the next one smarter.
Observe, predict, optimise, execute, learn. Five stages that run without a human in them, on every market interval, for every participant.
Grid and market signals ingested in real time, alongside forecasted weather data such as solar irradiance.
Probabilistic forecasts of demand, generation and price — distributions, so risk is priced rather than assumed away.
Forecasts weighed against market rules, asset constraints and risk limits to select bid, price and schedule.
Bids placed and dispatch issued autonomously, at machine speed, inside the interval that matters.
Outcomes retrain the models and feed straight back into Observe. The loop closes.
Four things we predict.
Consumption from household level up to feeder level.
Solar output, driven by weather conditions at the specific site.
Market clearing across intervals and products.
Full distributions rather than point estimates. A forecast without a confidence interval is an opinion.
Forecasts become decisions.
The optimiser weighs every forecast against market rules, asset constraints and risk limits, then selects the bid, price and schedule that maximises value for that participant.
What a trading desk does in hours, it does every market interval, for every participant at once. That is the only way this scales to millions of them.
Every trade that clears is a labelled example. The models improve on data a new entrant does not have and cannot buy.
An advantage that gets stronger over time.
Each new market reuses the same forecasting and optimisation core, configured for that market's rules.
Forecasting and optimisation models tuned against how these markets actually behave, not textbook assumptions.
Forecasting, bidding and settlement run in one system, so nothing is lost moving between steps.
Every trade improves the models. A competitor starting today has none of that data behind them.
Building the quantitative intelligence layer for future electricity markets.
Energy markets are being rebuilt around software, and most of that software does not exist yet. Founded in 2025, Quant Power is focused on developing this software.
Electricity used to be a few big decisions made slowly. It is becoming millions of small ones made fast, and the tools most of the market still relies on were built for the slower version.
We think the grid that emerges will be run by algorithms, and that whoever builds the good ones will matter enormously. So we are building them: forecasting, optimisation and execution as one platform, starting with the market that exists today.
We work with DISCOMs, microgrid operators, solar integrators and large energy users. We fit into existing tools and workflows rather than asking anyone to start from scratch.
Who's building it.
Mohan is a technology entrepreneur with proven experience in founding, building and exiting multiple startups.
He graduated in electrical engineering in 1983, followed by a masters in computer science in 1985, both from IIT Madras. He began his career in 1985 as a software engineer at CMC Ltd, and later joined Texas Instruments at Bangalore as one of their first set of employees in India.
His startups include ventures in Indian language processing (1989), embedded storage systems (1994), residential gateways (1999) and home security and automation (2009).
In 2004, Mohan re-joined Texas Instruments to handle their broadband and DSP business development in India, and established their place as a premier DSL and wireless infrastructure chipset supplier in India.
The last company, Silvan Innovation Labs, was acquired by Polycab Ltd in 2021.
Rajkumar is a telecom/wireless systems and embedded systems architect with 30+ years of experience in designing complex, large scale, highly-available systems. He has been involved in designing and implementing protocol stacks for telecom and embedded systems. He has developed core software for C-DOT's 40,000 line exchange, written software for DSL gateways, been involved in DSL modems for 2 wire line topology as well as home solutions including wireless security systems gateway, ATM security gateways and street light LED controllers.
He has worked with global MNCs like Lucent technologies, Texas Instruments and Cisco systems and homegrown startups such as Silvan Innovation Labs.
Rajkumar has an engineering degree in Computer science from Government College of Technology, Bharathiyar University.
Ashwin is an AI/ML architect and technical leader with 14 years of experience, including around 10 years in applied AI/ML. His work spans GenAI, agentic AI, multimodal systems, NLP, computer vision, forecasting, and optimization, with a focus on designing and building production AI systems end to end.
His quantitative work includes reinforcement learning, probabilistic forecasting, stochastic MPC, LP/MILP optimization, uncertainty-aware modelling, and energy-market and battery optimization. His background also includes graduate study in AI at Stanford and an M.S. in Electrical Engineering from North Carolina State University.
More recently, he has worked on LLM governance and routing, semantic reuse, RAG and agentic systems, multimodal AI, and model distillation, combining mathematical modelling, machine learning, system architecture, and product engineering.
Anirudh has built and led high impact R&D teams for Texas Instruments, Tektronix, Symbian, Nokia, Ixia Networks and Keysight. He was responsible for setting up the engineering centre in India for SiNett semiconductors, a tier-1 VC funded company. As VP/GM he has been responsible for end-to-end product delivery, innovation and growing sales in Indian/Asian markets. His focus has been in high technology areas like semiconductors, embedded systems, smart phones, networking and AI data centres. He has built teams and products from grounds up including market & product definition.
What changes if this works.
Prosumers earn more for surplus power; consumers pay less for green energy. Both come from a more efficient market, not from a subsidy.
Local trading reduces peak stress, network losses and costly redispatch.
Better price signals for solar accelerate rooftop adoption and cut curtailment.
Deep-tech quantitative infrastructure, built to extend to any liberalising electricity market.
Join a lean, algorithm-first team.
Our work is forecasting how energy is used and produced, and turning those forecasts into trades. On a team this small you sit close to both the models and the decisions they power.
There is no layer between the model you train and the market it trades in. You will see your forecast clear an interval, or fail to.
We are working on problems nobody has a settled answer to, with real money on the outcome. Both halves matter.
We like people who turn complex systems into something elegant, and who can explain it to a DISCOM engineer.
Time series, probabilistic models, weather-driven generation. Remote-friendly.
Constrained optimisation, clearing mechanisms, bidding strategy. Energy background welcome, not required.
Streaming data, metering integrations, execution systems that cannot miss a gate closure.
If you're good and the problem interests you, write anyway. We read everything.
info@quant-power.com →Let's talk about a pilot.
A feeder, a few hundred prosumers and a willing DISCOM is enough to start. Tell us what you have and we'll tell you honestly whether it's a fit.
We're also happy to talk to investors, researchers and anyone building on open energy infrastructure.