Laksh Gupta
Four years building backend systems that hold up in production — real-time services, AI platforms, and the infrastructure that keeps them shipping.
Systems
Four production systems across three different problem domains — AI retrieval, real-time distributed services, and infrastructure automation. Switch between them.
Brand Intelligence Platform
An end-to-end platform that ingests a brand’s web and social footprint and makes it semantically queryable as “Brand DNA”. I owned the full vertical — data model, ingestion concurrency, embedding pipeline, retrieval layer and observability.
- Designed domain-aligned PostgreSQL schemas and SQLAlchemy models first, so the ingestion layer had something coherent to write into.
- Built concurrency with AsyncIO and made runs resumable rather than restartable — a mid-run failure doesn’t discard completed work.
- Benchmarked Firecrawl, Apify, Selenium and Playwright per-source rather than standardising on one, because each fails differently under platform API constraints.
- Embedded via Vertex AI in quota-aware batches; vectors land in pgvector beside relational data so a single query can filter on both.
Brand DNA pipeline
A closer look at the hardest system I’ve built. Click any stage to see the engineering decision behind it — these are the trade-offs, not the tech list.
Sourcesweb · social
A brand’s website plus Instagram and X. The hard part isn’t reading them — it’s that every source rate-limits differently, changes markup without notice, and requires its own platform verification.
Capabilities
Six areas I've owned end to end, and where each was earned. Most were learned by being the person responsible when they broke.
REST and WebSocket services in Python and Golang — schema through versioning to auth. Concurrent request handling with AsyncIO, and real-time bidirectional messaging over persistent connections.
Domain-aligned relational schemas, SQLAlchemy models, ER diagrams before implementation. Restructured an unstructured data layer into something a team could extend without breaking it.
Agent workflows on LangGraph and the OpenAI SDK, RAG pipelines, vector search over pgvector with Vertex AI embeddings — plus the unglamorous half: quota batching, resumable runs, failure isolation.
Containerized microservices on Docker and Kubernetes, provisioning automated with Ansible, CI/CD through Jenkins, and automated product signing for verified releases to client infrastructure.
Run-ID structured logging, lifecycle monitoring, metrics instrumentation — built because I was the one debugging it at 2am. Directly cut support ticket volume by nearly a third.
RBAC across enterprise tooling, SAML single sign-on, security feature toggles, and pre-deployment test gates for regulated client environments.
Experience
- Architected an AI-powered Brand Intelligence platform end to end, working directly with founders.
- Shipped 20+ REST endpoints and autonomous agent workflows with Python, LangChain and the OpenAI SDK.
- Restructured an unstructured data layer into domain-aligned PostgreSQL schemas and SQLAlchemy models, authoring the ER and architecture diagrams the team now builds against.
- Built concurrent ingestion with AsyncIO and resumable run-lifecycle management across third-party platform APIs.
- Instrumented run-ID structured logging and lifecycle monitoring so every pipeline execution is traceable in production.
- Engineered LLM-assisted Playwright automation validating real user journeys, saving 8+ QA hours per release.
- Built a skill-centric AI QA agent with Python and Claude for structured knowledge retrieval.
- Contributed Golang microservices to an enterprise payment system, and produced system architecture and ER designs for a client VR platform.
- Practiced TDD and feature-toggle-gated deployment to client infrastructure.
- Engineered Golang REST endpoints and WebSocket services delivering real-time messaging to 15+ enterprise clients with persistent connection handling.
- Instrumented monitoring and metrics across distributed automation services, driving a 30% reduction in support tickets and 10% increase in adoption.
- Built RBAC and SAML-authenticated Django/MongoDB tooling; containerized microservices with Docker and Kubernetes, cutting deployment time 10%.
- Automated provisioning with Ansible and Jenkins CI/CD — 20% efficiency gain, 20+ engineering hours saved weekly.
- Implemented pre-deployment test coverage, security feature toggles and automated product signing for reliable client releases.
- Developed an RBAC-enabled MongoDB management UI with Python and Django.
- Built a Chart.js COVID-19 analytics tracker serving 1000+ global users, and cut deployment time 20% with Jenkins CI/CD.
Stack
Python · Golang · Java · Kotlin · JavaScript · SQL
FastAPI · Django · REST · AsyncIO · WebSockets · SQLAlchemy · Microservices
OpenAI SDK · Claude · LangChain · LangGraph · RAG · MCP · pgvector · Vertex AI
PostgreSQL · MongoDB · Redis · MySQL
AWS · Docker · Kubernetes · Jenkins · Ansible · CI/CD
TDD · Feature toggles · ER modeling · Structured logging · RBAC · SAML
Open to backend, platform
& AI infrastructure roles.
Bengaluru, India · open to remote and relocation