Manas Nikam

Updated on

Case studies

I'm Manas Nikam. What the client needed, what I built, what changed, and what I would do differently. Client work first, then hackathon builds, each labelled; the first one took a supplier booking from about 10 minutes to about 30 seconds.

Client work

10 min to 30 sper supplier booking, JonView

Client work

Integrating VIA Rail and JonView into a tour operator's booking system

Three travel suppliers wired into a live itinerary system for Fresh Tracks Canada: JonView search and booking with a circuit breaker and state sync, VIA Rail built end to end in five dated steps, a Rocky Mountaineer scraper. Per-product booking time went from about 10 minutes to about 30 seconds.

Stack Python, Django, Django REST Framework, PostgreSQL, Redis, WebSockets, AWS ECS, GitHub Actions, Salesforce, SonarQube

10 instanceshard cap on Vertex OCR; cost logged per invoice

Client work

Gemini invoice OCR and automated Tally invoicing for a printer-fleet business

Live for CD Infoware, an Indian printer-services company: invoice OCR on Gemini through Vertex AI with a 10-instance cap and per-invoice cost in USD and INR, plus a nightly Konica Minolta meter scrape that drafts one Tally-format GST invoice per device per month. 111 commits, all mine.

Stack TypeScript, React 19, Vite, Tailwind, React Native, Firebase, Vertex AI, Gemini, Puppeteer, Playwright, pnpm, Turborepo

11 metricsspec approved 23 July 2026, 30 of 30 commits mine

Client work

Monitoring a Postgres fleet with Lambda, DynamoDB, and one secret

Spec approved on 23 July 2026, then built and documented alone inside Fresh Tracks Canada's AWS account: EventBridge every 30 minutes, a collector Lambda in private subnets, DynamoDB with a 90-day TTL, a dashboard behind an internal ALB, 11 metrics with thresholds, and a least-privilege pg_monitor role. Adding a server is a secret edit, not a deploy.

Stack Python, PostgreSQL, AWS Lambda, Amazon EventBridge, Amazon DynamoDB, Application Load Balancer, AWS Secrets Manager, GitHub Actions

Hackathon builds

95% CIN runs per site, 8-category failure taxonomy, 13 WebMCP tools

Hackathon build

AgentReady: scoring whether the web is ready for agents

A benchmark that scores sites 0 to 100 on whether AI agents can complete tasks through the raw UI versus WebMCP tools: 13 WebMCP tools, serverless audits on Cloudflare (Queues, Browser Rendering, Workers AI, D1), an eval runner with N runs per site, 95% confidence intervals and an 8-category failure taxonomy. Built and submitted; outcome not recorded.

Stack TypeScript, Cloudflare Workers, vinext, D1, Workers AI, Browser Rendering, Cloudflare Queues, Playwright, WebMCP

3 in 3 daysagents on Cloud Run, 2 over vendor MCP servers

Hackathon build

Three agents on Cloud Run in three days, two integrated with vendor MCP servers

3 production agents on Cloud Run in 3 days, 2 integrated with vendor MCP servers (ClickHouse, Grafana): a screenplay clearance pipeline, a retention analyst over mcp-clickhouse, and a VFX war-room agent over mcp-grafana that writes back annotations and incidents. Each live with a demo video. Built and submitted; outcome not recorded.

Stack Python, Google ADK, Gemini 2.5, Vertex AI, FastAPI, Cloud Run, MCP, ClickHouse, Grafana, Prometheus, Loki, OpenTelemetry

Hackathon builds are where I test new tooling on a deadline; client builds are where it earns its keep; both are on this site, labelled, and I do not claim a placement I do not have.