I buildwebsiteswebsites, web apps, backends, platformsthat hold up.
I'm a software engineer in Hyderabad, currently at Deccan AI. I build websites, web apps and the backends behind them: APIs, queues and clusters that stay correct under load.
Next.js and React on the front, Go, Python and Java behind it. Open to full-time, contract and freelance work.
01Deccan Vault
View project- This is Deccan Vault. I built the whole thing, front to back.
- Teams browse, filter and open datasets, every screen mine.
- Behind it: AWS Lambda packaging, Redis rate limits, Kafka pipelines.

sample dataBuilt end to end
- ✓Website
- ✓Frontend
- ✓Backend & APIs
Under the hood
Dataset packagingAWS Lambda
Rate limitingRedis
Eval pipelinesKafka
Catalog storePostgreSQL
02Keploy
View project- At Keploy I taught tests to fake a database.
- It reads MongoDB's wire protocol straight from raw bytes.
- Then replays live Kubernetes traffic across seven protocols.
$ keploy test -c "go run main.go"
database: mocked · no mongod running
✓ pass GET /users/42
tests 5 · mocks used 18 · flaky 0
Record
No database needed
03CostObs
View project- This is CostObs. It shows what every LLM request actually costs.
- Drop-in SDKs for Python, TypeScript and Go. Zero added latency.
- Go and ClickHouse answer cost breakdowns in under a second.
LLM spend · this month
$4,812.40
by modelby customerby feature
Top models
gpt-4o1.9M$2,140
claude-sonnet1.2M$1,486
gemini-flash3.4M$612
embeddings8.0M$574
pythontypescriptgo
from costobs import track client = track(OpenAI(), customer="acme-corp", feature="support-chat")
+0 ms on the request path
One request
- model
- gpt-4o
- tokens
- 812 → 264
- customer
- acme-corp
- cost
- $0.0047
SELECT feature, sum(cost)
FROM events WHERE tenant = ?
GROUP BY feature
4 rows · 9.5M events · 312 ms
What I build
fiHire me on Fiverr↗- WebsitesPortfolios, landing pages and marketing sites. Fast, accessible, and they move.
- Web apps & MVPsFull-stack products end to end: auth, dashboards, payments, deploys.
- Backends & APIsAPIs, queues, workers and integrations that stay correct under load.
- Cloud & DevOpsDocker and Kubernetes on AWS or GCP, with Prometheus and Grafana watching.
Experience
Click any role for the details.

Python
NumPy
pytest
GitHub Actions
2× Google Summer of Code
LF Mentorship
ICPC Regionals
Got a project?
I'm open to full-time, contract and freelance work.
officialasishkumar@gmail.com


