Skip to content
JS

Case studies

Five systems, and what each one taught

Written for someone who will have to work with me. Each page states the constraint, shows the architecture, names what I owned, and ends with what I would change.

Autonomous incident investigation on LangGraph

Apple · Oct 2025 — Present

An SRE that reads the runbook before it pages you

A multi-agent platform that takes a production alert, retrieves grounded evidence, proposes a root cause with citations, and asks a human before it touches anything.

reduction in MTTR

30%

reduction in MTTR

alerts triaged per day

10k+

alerts triaged per day

unattended production writes

0

unattended production writes

LangGraphPythonRAGKnowledge GraphsAWS

Streaming + batch market data on AWS

Block Scholes · Sep 2024 — Sep 2025

Ten million ticks a day, and a storage bill that stopped growing

A Kinesis-to-S3 tick pipeline with a compaction job behind it, because the thing that kills a Parquet lake is not volume — it is object count.

records processed daily

10M+

records processed daily

storage cost reduction

40%

storage cost reduction

test coverage

+30%

test coverage

AWS KinesisS3PySparkParquetAthena

AML surveillance over streaming behavioural signals

TikTok Live (ByteDance) · Dec 2022 — Dec 2023

Watching seven million live rooms for the wrong kind of money

Risk-detection models applied to live behavioural streams, plus the creator analytics platform underneath it, served at 350K+ QPS.

live rooms monitored daily

7M+

live rooms monitored daily

QPS served

350K+

QPS served

read-latency reduction

65%

read-latency reduction

GoKafkaRedisRisk ModelsStreaming

Firmwide observability and batch orchestration

Goldman Sachs · Jan 2018 — Nov 2022

Eleven million queries a day, and the alerting that came with it

An Elasticsearch-backed observability platform powering firmwide search and alerting, and the orchestration layer governing ten thousand scheduled jobs behind it.

queries indexed daily

11M+

queries indexed daily

scheduled jobs governed

10k+

scheduled jobs governed

production incidents

~0

production incidents

JavaElasticsearchAWSSpring BootKibana

M.S. by Research — neural information retrieval

IIIT Hyderabad · May 2016 — Aug 2018

Teaching a retriever what a bug report is really about

Dense retrieval for short, noisy documents: a BERT model trained with triplet learning and inter-document context, and a screening system that saved a third of reviewer time at full recall.

screening time saved at 100% recall

31%

screening time saved at 100% recall

Recall@1 across three corpora

61–65%

Recall@1 across three corpora

recall improvement over baselines

10–15%

recall improvement over baselines

BERTPyTorchDense RetrievalTriplet Learning