Serentica Energy Intelligence Platform
India's renewable energy operators manage real-time power markets across fragmented spreadsheets and portals. This is the design story of replacing that chaos with one cohesive system.
Role
LEAD UX DESIGNER
Duration
3 months
May 2025 to Aug 2025
Platform
Web Appplication
B2B SaaS · Internal tool
Modules
5 shipped
End-to-end owned
disclaimer
This project is covered under a Non-Disclosure Agreement. Only select screens and non-sensitive design artifacts are shared here. Proprietary algorithms, client data, and internal tooling have been omitted. The work shown is representative of the overall UX approach and design system.
01
CONTEXT
A market built on
milliseconds and megawatts
India's IEX market runs three trading windows — GDAM, DAM, and RTM — each with its own bidding logic and settlement cycle. Solar generation varies every 15 minutes. Miss a window by an hour and you're leaving significant revenue on the table.
Serentica operates utility-scale SPVs across India. Before this platform, the team ran everything through spreadsheets, email reports from three vendors, and manual grid submissions.
Key risk —
"By the time I compile forecasts from three vendors and check prices on IEX, the market window is already closing."
— Operations Planner, site visit
"We know we're losing money in DSM calculations, but we can't see where until the monthly report — by then it's too late."
— Senior Energy Trader, discovery interview
02
PROBLEM
Operators were flying
blind in real-time
Through discovery sessions, we identified a system where critical decisions were made on outdated information, siloed tools, and gut instinct.
No unified forecast view
Operators toggled between three vendor portals, a government grid portal, and Excel sheets to piece together a picture.
Revenue leakage invisible
DSM penalties only surfaced during monthly reconciliation. Financial impact of daily decisions was never visible in real-time.
Manual bid optimization
Power allocation across GNA and TGNA was done manually in Excel — no algorithm, no scenario modeling.
No market timing alerts
DAM closes 12+ hours ahead. RTM runs on 30-minute blocks. Zero proactive reminders for these critical deadlines.
HOW MIGHT WE
Help energy operators make faster, more accurate market decisions — without ever leaving the platform?
03
USERS
Two user types,
one connected system
Operations Planner
Day-to-day market ops
O
Can do
Monitor real-time generation vs. forecast
Review prices and submit bids
Track DSM exposure live
Select best forecaster per block
Cannot do
Modify platform configuration
Energy Trader
Revenue strategy & optimization
P
Can do
Analyze revenue across market windows
Run What-If scenario simulations
Access consolidated multi-date analysis
Compare Base vs. Optimised vs. Peak
Cannot do
Edit bid allocations or forecasts directly
04
RESEARCH
Understanding the
operator's mental model
Week 1–2
Contextual Inquiry
Shadowed 6 operators live during morning market sessions — full workflow from forecast to bid.
Week 2–3
Stakeholder Interviews
18 structured interviews — operations, finance, trading desk, dispatch engineers.
Week 3–4
Data Analysis
6 months of bid data, DSM penalty logs, and forecaster accuracy records analyzed.
Week 4–5
Synthesis
Affinity mapping + HMWs translated into a prioritized backlog with product team.
"Operators weren't asking for more data — they were asking to stop seeing data that wasn't relevant to right now."
1
Time is the primary constraint
Every single operator named market timing as their #1 stress. Urgency needed to be a first-class design element — not an afterthought.
2
Forecaster trust is based on memory, not data
Operators had gut-feel about which vendor was "good" — but MAPE varied significantly by time-of-day. They needed live evidence, not instinct.
3
Revenue feedback loop was completely broken
No same-day financial feedback on decisions. The connection between a bid and its outcome was severed — wouldn't surface for 30 days.
05
INFORMATION ARCHITECTURE
Structuring complexity
into a clear mental model
The platform's IA was designed around the operator's decision workflow — not the data's technical hierarchy.
⚡Serentica Platform
Platform Navigation Architecture
Design decision:
06
DESIGN DECISIONS
Three options considered.
One reason each was chosen.
Decision 01 — Market Countdown
RULED OUT
Notification bell
Operators never checked the tray during high-focus sessions. Alerts went unseen.
RULED OUT
Static status text
"DAM closes at 10:00" in the header. Became invisible after day one — processed as metadata.
CHOSEN
Persistent live countdown
A live timer in the header, dynamically colored by urgency. Present on every single screen without fail.
Decision 02 — Forecaster Selection
RULED OUT
Monthly MAPE report
Historical data is useless when making a decision in the next 5 minutes.
RULED OUT
Side-by-side raw numbers
Required mental calculation to identify best — adding cognitive load at exactly the wrong moment.
CHOSEN
Live MAPE ranking
Current-day MAPE per forecaster, ranked in real-time. Best performer auto-highlighted. Zero calculation required.
Decision 03 — Alert Architecture
RULED OUT
Toast notifications
Auto-dismissed before operators noticed them. Critical alerts disappeared silently.
RULED OUT
Unified alert panel
Operators had to navigate away from their workflow — breaking focus entirely.
CHOSEN
Tiered inline alerts
General (platform-wide) + Specific (screen-level). Alerts appear inline where action is needed. No navigation.
Decision 04 — Revenue Display
RULED OUT
Single actual figure
₹3.78Cr in isolation means nothing. No baseline, no context, no signal to act on.
RULED OUT
Historical comparison only
Showed what happened last week — not what was possible today. Missing the action signal entirely.
CHOSEN
Base / Optimised / Peak trio
Three states simultaneously — what you earned, what the algorithm saw, what was achievable. The gap became the signal.
07
SCREENS
Selected screens
from the platform
Screen 01 — Power Market Prices
A — Dual chart
Forecast vs. Actual
Two charts stacked — Forecast top, Actual IEX below. Parallel layout for instant comparison without navigation.
B — Table sync
15-min granularity
Scrollable table beneath each chart. Hovering a time-slot highlights it in both chart and table simultaneously.
C — Multi-date
Average overlays
Toggle average prices from a second date range. Dotted lines = averages, solid = current day. Instantly readable.
Screen 02 —DSM & Revenue Analysis
A — Bid table
GDAM / DAM / RTM / TGNA
Bid performance by market segment. Color-coded labels consistent across the entire platform.
B — Revenue trio
Base, Optimised, Peak
Three revenue states side-by-side — the morning-check screen operators open first each day.
C — KPI gap
Optimisation
Negative gap in red = urgency without drill-down. Accuracy % confirms forecast quality. Both instant.
08
IMPACT
The MAP finally
made sense
Measured over 3 months post-launch, comparing the same planning workflows before and after platform adoption.
82%
Forecast accuracy via consolidated forecaster — up from ~61% on manual approach
~40%
Reduction in time from forecast receipt to final bid submission
₹Cr+
Revenue uplift through optimization features — exact figures withheld under NDA
"I can actually see the impact of my decisions on the same day now. Before this I was completely blind until the monthly report."
— Operations Planner, post-launch feedback
09
LEARNINGS
What this project
taught me
1
Domain fluency is a design prerequisite
I spent 1 weeks purely learning IEX market structure before opening Figma. Operators immediately trusted the interface because it spoke their language.
2
The workflow IS the IA
Early versions organized by data type. The breakthrough came when we restructured around the operator's actual morning sequence — not where data lived.
3
Numbers need anchors
The Optimisation gap became the most-discussed element in every user session. A number without a reference is just noise — relationships are what build confidence.
4
Real-time context changes the entire design language
The same screen looks different with 12 hours until close vs. 20 minutes. Adaptive urgency had to be baked into the system from the ground up.
5
Design system as shared contract with engineering
Defining what's realtime vs. daily vs. historical upfront eliminated every "the data doesn't work like that" conversation late in development.
More to this project than what's shown here
NDA limits what I can share publicly — but there's substantially more behind this. Multiple complex modules designed and shipped.
Generation Forecast Analysis
Optimised Power Allocations
What-If Simulator
Battery Optimisation
More Projects
Logistics
B2B
Web App
2025
FIFA Logistics Platform
Transportation planning tools for 500+ vehicles, dozens of venues, and hundreds of personnel across major FIFA events.
View case study
Ed-Tech
B2C
Mobile App
2021
Smart Class
Mobile learning platform for civil engineering and architecture students — video lessons, quizzes, and progress tracking for complex material on the go.
View case study
Healthcare
B2B
Web App
2026
Care Connect
Call tracking platform for healthcare front desk teams — inbound queue management, instant patient context, and zero missed follow-ups.
View case study