Introduction
Professional Journey
My Studies
What I'm Good At
Asia/Kolkata
English (B1)
German (A1)
Ayush Mania
Forward Deployed Engineer & Platform ArchitectI'm Ayush Mania, a forward deployed engineer and platform architect. I work embedded with product teams to solve problems where they happen - designing systems that help businesses scale without losing reliability. My focus spans backend engineering, cloud infrastructure, and applied LLM systems: semantic search, agentic workflows, and the unglamorous plumbing that keeps them honest in production. I turn complex problems into simple, efficient solutions that ship.
Professional Journey
TestDino Inc.Jan 2025 - Present
Forward Deployed Engineer & Product/Platform Architect- Founding engineer on TestDino, a SaaS platform for test reporting, observability, and test intelligence. Now serving 80+ organizations and 130+ engineers and QAs, with 2M+ test executions analysed, at 99.95%+ uptime across all services.
- Designed backend architecture from authentication to analytics across nine services in three runtimes: Go for event ingestion and the realtime analytics plane, Node.js and Express for the tenant, billing, and integration services, and Python and FastAPI for the ML service. Data sits on PostgreSQL 17 and TimescaleDB, with MongoDB Atlas behind the configuration services and Redis Streams carrying realtime fan-out.
- Built tdpw, the published Playwright test-data collector, so teams adopt TestDino with one npm install and a single line of reporter config rather than changing how they work.
- Managed DevOps pipelines and Azure-based infrastructure, improving deployment speed and uptime consistency.
- Integrated Customer.io for automated engagement flows, improving product feedback cycles.
- Grew the engineering org from one engineer to a backend team, writing most of the platform in the first four months and then shifting to architecture, review, and written decision records as the team scaled.
Darkark Technologies2026
Engineering Lead, Client Engagements (independent contract)- Diamond manufacturing operations platform. Owned end to end in 21 days. Floor staff record carats but must never reach cost, and the platform issues every signed-in user the same database role, so column grants could not separate owner from clerk. Revoked money columns from that role and re-exposed them through security-definer views carrying their own membership test. Cost follows weight through a cutting tree in exact decimal arithmetic, with billing rows append-only. 27 migrations with paired rollbacks, 54 row-level security policies, 42 constraint assertions, 11 decision records.
- Drivetrain TCO and route feasibility for an automotive client in Munich. Direct client interface over a six-week weekly feedback loop. Seven drivetrain concepts across consumption, total cost of ownership and European route feasibility, with a tapered charging model feeding a stop-placement solver. Modelled post-2031 German toll policy as user-selectable scenarios rather than a constant. 184 tests, bilingual German and English UI, ahead of schedule on every milestone.
- Saral, AI recruiting search (Headsin LLP). Owned platform architecture, infrastructure and the entire search data flow over 119k+ indexed profiles: a BGE-M3 embedding sidecar, hybrid vector plus metadata retrieval, deterministic hard scoring, then LLM post-scoring. Replaced a monolithic backfill with progressive mini-batches on pg-boss, cutting first results from 90-120s to 25-40s. Shipped an append-only credit ledger with idempotent per-page charging and forward refunds.
- B2B catalog and RFQ platform. Three deployed surfaces from one workspace in two weeks, with a shared identity package so the same product redeploys for another tenant without editing the apps. Two entry points from one API codebase, long-running and serverless, with cold-start requests funnelled onto a single connection handshake. 174 tests across unit, API integration and end-to-end.
- Took over second-phase engineering on the company site: audited it page by page, removed every unsourced quantitative claim, and wired a pre-write hook that blocks any edit reintroducing one.
Alphabin Technology ConsultingDec 2023 - Dec 2024
Backend Engineer- Designed and developed a Test Execution Engine to run thousands of test cases in parallel on cloud infrastructure with minimal cost and optimal speed.
- Built and managed the full stack using Terraform, Kubernetes, Docker, GitHub Actions, Azure, Lens, Node.js, MongoDB, and Redis for scalable, reliable performance.
- Worked with the content team to share insights on backend security, contributing to awareness blogs and internal documentation.
- Gained hands-on experience with Google Cloud Run and Cloud Buckets by deploying the Pytest runner, a core part of the larger system architecture.
Virtually Testing FoundationJul 2023 - Sep 2023
Red Team Engineer Intern- Acquired expertise in advanced penetration techniques, and worked on advanced projects marking a significant milestone in my professional journey and propelling me to advance in Cybersecurity.
My Studies
Nirma UniversityBachelor of Technology in Computer Science and Engineering, Jun 2020 - Oct 2024. Coursework in programming, algorithms, networking, computer systems, and security.
What I'm Good At
Platform ArchitectureI design SaaS platforms that stay stable, fast, and affordable even when they scale. My focus is on systems that work quietly but powerfully behind the scenes.
System Design
Scalability Planning
Cloud Infrastructure
Cost Optimization
Distributed Systems
Multi-Tenant SaaS
Backend & Full-stack EngineeringEnjoy building clean, dependable systems that handle real-world load without losing clarity or performance.
Node.js
TypeScript
Python
Fastify
Express
FastAPI
REST APIs
Microservices
Caching
Asynchronous Processing
Job Queues
PostgreSQL
TimescaleDB
MongoDB
Redis
Drizzle ORM
React
Next.js
Git
AI & Applied LLM EngineeringI build LLM features that hold up in production: bounded, observable, and honest about what they do not know. Degraded results carry a flag; they never fail silently.
Agentic Workflows
Semantic Search
Vector Databases (pgvector)
Embeddings (BGE-M3)
LLM Fallback Chains
Groq & Azure OpenAI
AI-Assisted Development
Claude Code
Cursor
Model Context Protocol (MCP)
DevOps & InfrastructureI'm comfortable with the full stack of infrastructure, from setting up CI/CD pipelines to deploying and monitoring systems in production.
Terraform
Kubernetes
Helm
Docker
Azure
AWS (ECS Fargate)
Google Cloud
GitHub Actions
Observability
Technical LeadershipI work with teams to turn ideas into working systems, making sure architecture decisions actually serve the product, not the other way around.
Forward Deployed Engineering
Product & Platform Architecture
Technical Roadmapping
Communication
Documentation
Code Review Strategy
Decision-Making