Things I’ve shipped

Built Caylex Backoffice from scratch — the internal staff console

One place for the team to onboard MCP servers, watch background jobs, flip feature flags, find and clean up orphaned data, and see how the product's being used.

Shipped Caylex's MCP onboarding pipeline

Checks each new MCP server, runs the whole onboarding on its own, and publishes only the ones that pass a health check. Shipped as a CLI and a skill: any server with self-registration (DCR) onboards hands-free.

Set up Caylex's internal observability

The OTel pipeline that pulls logs, metrics, and traces from across the system into one place — the backbone the LLM tracing plugged into a month later.

Shipped external observability for customers

Customers can route their own telemetry straight from Caylex to whatever monitoring tool they already run — plus the admin UI to set it up.

Added LLM observability

Tracing on every model call: prompt, tokens, latency, cost. When an agent misbehaves, we replay exactly what it saw. Built on the OTel pipeline I'd set up the month before.

Redesigned + shipped paramteraiya.me ↗

A new design system, built so AI agents can read it too (JSON-LD, llms.txt), with analytics. Live.

Integrated Skills into the Caylex platform — progressive context disclosure

Agents pull in a skill — its instructions and tools — only when a task needs it, not on every request. The library can grow large while any single agent's context stays small.

Shipped the Security Findings UI

Surfaces security issues found on connected servers and their tools, with a detail panel and advisory fixes, plus notifications when something needs attention.

Built the agent identity editor (Navigator)

The UI to give an agent its identity and instructions in plain markdown — its name, description, and how it should behave.

Built the Tool Permissions & Risk console

An admin sees how risky each tool is, sorts and filters by severity, bulk-disables the dangerous ones, and overrides a rating when needed — governance over what an agent is allowed to touch.

Built Context Maps for Caylex

A memory of what lives in each connected tool — Linear, Notion, Slack. The agent checks the map first, then queries only the tool that has the answer. I built the backend and the config UI.

Built an MCP server classifier (later folded into Caylex)

Works out what each server and its tools actually do, scores how much each tool can affect things, and flags security risks.

Built server discovery & connection for Caylex agents

The flow for browsing the library of available servers, filtering by category, seeing details, and connecting them to a project — plus session recovery and faster page loads.

Built deep-reasoning agents & tool-use pipelines at TheAgentic

Agents that reason through multi-step problems and call the right tools, wired into scalable systems for industry-specific workflows.

Shipped a Jira integration at TheAgentic — OAuth flow, project management, Linear/Jira mutual exclusivity

Took a vertical AI platform's frontend 0→1 at TheAgentic — releases, integrations, dashboards, launch commander

Built backend APIs at TheAgentic — dashboards, users, profiles, GitHub integration

Built laaganinaturecure.com — a nature-cure site with an interactive quiz + CTA optimization ↗

Built an MCP server that streams logs & traces to LLMs

An MCP server that pipes live logs and traces to the LLM — every debugging prompt arrives with real observability context, not just the question.

Featured in Analytics India Magazine — "Why Do Observability Platforms Even Exist?" ↗

Guest on the "AI IS HERE" / DEV BTS interview series

Shipped ML-based Log Clustering at Middleware

Improved root-cause-analysis speed by 60%, cutting incident resolution from hours to minutes for large-scale cloud environments.

Created a Root Cause Analysis AI agent

Traces a failure back to its source and proposes the code or config fix — the triage pass an on-call engineer would otherwise do by hand.

Led a 2-engineer AI team at Middleware — anomaly detection, text-to-query, forecasting

Delivered the log module — production APIs & frontends

Built the backend APIs and the UI for logs, the foundational pillar of an observability platform.

Built a data pipeline & scheduler framework for BloomChat AI

A Python framework that syncs third-party sources into BloomChat AI on a schedule, keeping its data fresh without manual imports.

Hardened data quality with Pydantic validation in FastAPI

Added Pydantic validation across the FastAPI endpoints, catching bad data and schema drift at the service boundary before it reached downstream pipelines.

Designed the AI engine + backend for live image sharing at Laracon India 2023

Node.js, PostgreSQL, a vector DB, and AWS powering real-time image sharing for attendees at Laracon India 2023.

Built a high-throughput backend handling millions of events/hour

Kafka, FastAPI, and MongoDB on Azure, processing millions of live-stream events every hour.

Built cross-platform worker apps in Go & Python

Automated model training and data exchange across distributed nodes.

Built & deployed furniture recognition models on AWS

Deep-learning models that recognize furniture, trained and deployed on AWS.

Developed hyperspectral image reconstruction from RGB

Algorithms that rebuild full hyperspectral images from ordinary RGB input.

Built IoT sensor pipelines for real-time factory monitoring

Integrated sensor data pipelines for live process monitoring on the factory floor, plus IoT-based permeability and viscosity tests using microcontrollers.

Built an ANPR system with OpenCV ↗

Automatic number-plate recognition in Python — detecting and reading vehicle plates from images with OpenCV.

Built a data-storytelling project ↗

Turning raw data into a clear narrative — a storytelling project (Alpha AI).