Abhijit.

A reconciliation platform whose matching logic pairs transaction records across three independent sources and flags every mismatch across lakhs of rows.

• Matching logic pairs records across the ATM switch, ATM electronic journals and the core banking system, flagging every mismatch across 100,000+ rows.
• Python parsers decode binary-format machine files into structured records — turning raw data into input that can be matched and audited at scale.

rows per run

100K+

independent sources

3

Generates the bills banks such as Axis and SBI owe for cash management and cash van services, across multiple states, on a schedule and without manual effort.

• Bill generation runs on Celery workers and Python cron jobs, so billing cycles run on a schedule instead of by hand.
• Retries and idempotency mean a re-run never double-bills. Bills are generated for banks like Axis and SBI across multiple states.

banks billed

2+

billing services

2

A bond custodian platform that safekeeps bonds for investors and settles their trades, delivered by a small cross-functional team spread across three countries.

• Object-oriented C#/.NET REST services implementing SWIFT MT and MX messaging so transaction records met regulatory reporting obligations in Brazil and South Korea.
• Supported millions of BRL in daily settlements.

interactive screens

33+

countries, one team

3

An infrastructure-intelligence platform over 25M+ records, with an AI agent that plans a question, calls the right tools and comes back with an answer.

• Built an AI agent that decomposes a question, calls search, filter and summarise tools in sequence, and assembles the result — rather than answering from a single prompt.
• Diagnosed Elasticsearch bottlenecks and designed multi-tier caching: API latency 5s to 0.8s, infrastructure spend down $3,200 a month.

records searched

25M+

to an insight

4h → 3min

enterprise clients

50+