POCKET-SIZED
01 / about
Software with the
thinking left in.
This portfolio connects the systems I build to the engineering decisions, experiments, and lessons behind them.
About me
I’m Namah, a Senior Software Engineer with 8+ years of experience designing and scaling backend systems, distributed data platforms, and enterprise APIs.
My work focuses on performance engineering, concurrency, reliability, Kubernetes, and backend architecture—from systems handling ~100K RPS to platforms processing 125M+ daily events and petabyte-scale datasets.
About Puhtaeto
Puhtaeto is my collection of pocket-sized software and honest engineering stories: what a system does, how it works, the trade-offs involved, and the evidence behind it.
Currently building in public
02 / featured build
Browseterm.
A real system, documented from product idea through distributed architecture, security, reliability, and operations.
BUILD 001 · 2023—PRESENT
BrowseTerm
A bring-your-own-machine development platform that lets users create, save, hibernate, resume, and open isolated Linux workspaces from a web browser while workloads run securely on their own computer.
- Architected a hybrid cloud platform deployed on Contabo k3s with Kubernetes, Traefik, cert-manager, and Cloudflare Pages.
- Built a Python desktop/CLI runtime using Multipass and k3s, with an outbound TLS bidirectional gRPC/Protobuf stream to the cloud.
- Kept the user’s Kubernetes API private while the Device Agent communicated with Container Maker over mTLS.
- Engineered reliable CREATE, DELETE, SAVE, HIBERNATE, and RESUME workflows with command queues, execution journals, idempotent redelivery, atomic quota reservations, and placement-generation checks.
- Implemented secure browser terminals with Node.js, WSS, SSH, ngrok, and single-use tickets; added immutable workspace snapshots and centralized logs with Grafana, Loki, and Alloy.
PYTHONFASTAPIJINJA2POSTGRESQLSQLALCHEMYALEMBICREDISSQLITEOAUTH2RESTSSEGRPCPROTOBUFMTLSMULTIPASSK3SKUBERNETESTRAEFIKCERT-MANAGERNODE.JSWEBSOCKETSSSHNGROKMINIODOCKER HUBGRAFANALOKIALLOYCLOUDFLARE PAGES
BUILD 002 · IN PROGRESS
iKompare
An AI-powered price comparison platform designed to help users find and compare prices for any product across countries and retailers through a simple natural-language enquiry.
In progressLive site · Soon
- Building a multi-agent price finder to discover products, identify relevant variants, and research retailer offers.
- Matching listings by model, specifications, and condition to make comparisons accurate and meaningful.
- Bringing together prices, availability, and purchase links in one place, with sources and timestamps for transparency.
- Enabling cross-country comparisons with currency conversion to help users discover better deals.
03 / engineering journey
Experience, as a trail.
Eight-plus years across backend systems, distributed data, infrastructure, developer platforms, and performance engineering.
Enterprise APIs, execution systems, and performance
- Solely designed and scaled an enterprise external API platform to ~2,000 RPS at ~195 ms p99 on a single-core CPU in load testing.
- Implemented OAuth2 Client Credentials with Ory Hydra and built a reusable NetworkX-based JSON-to-SQL query builder.
- Reduced critical transformation latency by ~80% and eliminated gateway timeouts using concurrency, caching, Kafka, and asynchronous execution.
TypeScriptNestJSPythonFastAPIOAuth2KafkaMongoDBClickHouseKubernetesAWS S3
Petabyte-scale seismic data for TGS Energy & Data
- Processed petabyte-scale seismic datasets with Python, Dask, Zarr, and MDIO.
- Migrated synchronous I/O to AsyncIO and benchmarked caching against latency, storage requests, and worker memory to avoid unnecessary production complexity.
PythonAsyncIODaskZarrMDIODistributed Computing
Extensible healthcare load-testing platform
- Led 4–5 engineers to redesign the load-testing platform into an extensible Python/FastAPI application.
- Improved concurrent workloads with AsyncIO, background tasks, and ThreadPoolExecutor; deployed distributed load generation through Kubernetes and GitLab CI/CD.
PythonFastAPIAsyncIOThreadPoolExecutorDockerKubernetesGitLab CI/CD
Multi-vendor network automation
- Built Python integrations using CLI communication and TextFSM to convert unstructured router output into structured platform data.
PythonTextFSMNetwork AutomationCLI
Healthcare IoT pipelines and observability
- Redesigned a failing IoT pipeline to process ~5,000 messages/minute with Kafka and PySpark, including late-arriving events and standardization.
- Correlated Kafka lag and throughput with CPU/memory saturation using Prometheus, Grafana, and PromQL to tune Spark capacity and batch sizes.
- Independently built a Python load tester and PySpark application-generator CLI adopted by engineering teams.
PythonPySparkKafkaKubernetesFissionTimescaleDBPrometheusGrafana
Browser-based 3D interior design
- Built the initial Three.js interior-design experience and extended the Python/Flask backend with APIs and data-model changes.
PythonFlaskJavaScriptThree.jsREST APIs
Recommendation APIs and subscription analytics
- Maintained recommendation APIs handling ~100K RPS and used CloudFront to offload peaks of ~300K requests/sec.
- Optimized ~400 ms latency through MongoDB tuning, Redis caching, and ThreadPool concurrency.
- Solely built a PySpark platform ingesting ~125M events daily, reducing analytics API latency from ~8 minutes to ~15 ms.
PythonFlaskPySparkCassandraMongoDBRedisCeleryCloudFrontKubernetes
End-to-end IoT street-lighting platform
- Sole engineer across cloud infrastructure, backend APIs, Raspberry Pi edge software, and web management.
- Built remote monitoring and control with AWS IoT Core, MQTT, API Gateway, Lambda, Cognito, Python, and JavaScript.
PythonJavaScriptAWS IoTMQTTLambdaCognitoRaspberry Pi