// selected work

Real clients. Real problems.
The metric we moved.

Not a portfolio of screenshots. Every case below is a named client (or anonymised by request), a specific problem, and the measurable we actually shifted.

150+
Projects shipped
6
Countries served
0
Missed deadlines, 2024
3+ yrs
Median client tenure
Filter:Showing 8 of 8
/01E-commerce

From 400 to 12,000 daily orders

A Shopify store outgrowing the platform, rebuilt headless on Medusa with a 2-tap UPI checkout and automated COD-OTP.

Challenge

Checkout broke on 2G, COD confirmation was manual, Black Friday would have crashed the stack.

D2C home essentials
Bangalore · 14-week rebuild · 14 weeks
30×
Daily orders
41%
Checkout conv.
99.4%
Payment success
2.1s
Mobile LCP
Stack
MedusaNext.js CommerceRazorpayShiprocketMoEngageAWS
Read the service page
/02Real Estate

Follow-up rate 23% → 91%

Custom CRM + WhatsApp orchestration that made non-follow-up impossible.

Challenge

77% of leads died silently in an agent's WhatsApp. No SLA, no escalation, no visibility.

Brokerage · Pune
12 brokers · 400 leads/mo · 11 weeks
91%
Follow-up rate
3.2×
Close ratio
₹2.4Cr
Added GMV/yr
18%
Site-visit conv.
Stack
Next.jsNode.jsPostgreSQLWhatsApp Cloud APIExotelRedis
Read the service page
/03Manufacturing

Month-close 11 days → 2 days

Unified ERP replacing Tally + 4 Excel files at a 140-person FMCG company.

Challenge

Inventory across plants lived in four spreadsheets; month-close took two weeks of reconciliation.

FMCG manufacturer · Coimbatore
140 people · 3 warehouses · 20 weeks
2 days
Month-close
99.4%
Inventory accuracy
₹0
GST penalties, 12m
40%
Ops time saved
Stack
Next.jsNestJSPostgreSQLTally ConnectorGSTN APIsClickHouse
Read the service page
/04Fintech

₹12 Cr disbursed in 6 months

RBI-compliant digital lending platform from in-principle licence to production. Audit passed first try.

Challenge

RBI timelines meant the MVP had to be production-grade, not a prototype.

NBFC · Chennai
RBI-registered · in-principle · 11 weeks
₹12Cr
Disbursed
11 weeks
To launch
0
Audit findings
100%
RBI compliance
Stack
Next.jsNode.jsPostgreSQLAccount AggregatorNACHAWS KMS
Read the service page
/05Healthcare

0 → 50K DAU · featured by Apple

iOS-first app for affluent urban users booking premium home care. Featured in "New Apps We Love".

Challenge

The target audience expects App Store-featured quality from day one. No room for "the Android version first" compromise.

Healthtech startup · Bangalore
Apple Featured · 14 weeks
4.9★
App Store rating
50K
Daily users
22%
D30 retention
1
Apple feature
Stack
SwiftSwiftUICombineCloudKitXcode CloudRevenueCat
Read the service page
/06SaaS

Tamil NLP at 96% accuracy

Content classification across 2.4M Tamil articles — entity extraction, sentiment, auto-taxonomy.

Challenge

Generic multilingual models butcher Tamil morphology. Google Translate was getting it embarrassingly wrong.

Tamil media group · Chennai
Decades-old archive · 12 weeks
96%
Classification
2.4M
Articles processed
14
Content taxonomies
−38%
Publishing time
Stack
IndicBERTMuRILHuggingFaceLabel StudioFastAPIPinecone
Read the service page
/07Fintech

80% tier-1 automation in 4 languages

Multilingual chatbot handling loan-status, EMI, payment, documents in Tamil / Hindi / Telugu / English.

Challenge

Tamil-speaking customers were hanging up on call-centre agents who couldn't understand them.

Financial services · Chennai
Loan + EMI support · 10 weeks
80%
Tier-1 automation
4.6/5
CSAT
−34%
Support cost
1.4s
Median response
Stack
OpenAI GPT-4oLangChainPineconeWhatsApp Cloud APITwilioPostgreSQL
Read the service page
/08Logistics

Error rate 2.3% → 0.04%

Rewrote a legacy REST layer into a versioned, documented, idempotent event-driven API with full observability.

Challenge

Four different "APIs" held together by assumptions and WhatsApp messages. P99 latency was 3.2 seconds.

Marketplace · Mumbai
14M+ API calls/day · 16 weeks
14M/d
Requests handled
190ms
P99 latency
0.04%
Error rate
99.99%
Uptime
Stack
Node.jsFastifyKafkaPostgreSQLRedisDatadog
Read the service page
// want yours here?

Let's make the next case study about you.

The cases above all started with a 30-minute call. Yours will too.