Hi, I’mAditya
Dhanraj
I build high-performance APIs, developer platforms, and backend systems that scale.
About
I'm a backend engineer currently at Infineon Technologies, where I own the API platform and define standards across the org. My work sits at the intersection of software engineering, DevSecOps, and developer tooling. I care about systems that are fast, observable, and maintainable at scale. I'm not a framework tourist; I pick tools that solve problems. My stack leans backend-first: .NET, Node.js, Python, Postgres, Redis, Docker, Kubernetes. I'm equally fluent in cloud-native CI/CD and AI agent orchestration. Outside work, I'm building side projects that push production limits and grinding competitive programming to stay sharp.
Year at Infineon
API Latency Reduction
Problems Solved
Experience
Software Development Engineer — API Platform
- Developed an internal registry for reusable AI agents and developer tools, accelerating adoption of AI-assisted workflows across engineering teams.
- Define API standards, test coverage requirements, and automated quality gates enforced via CI/CD pipelines.
- Lead API governance initiatives — driving consistency, versioning, and discoverability across 30+ internal services.
- Building the next-gen developer platform that abstracts infrastructure complexity for backend teams.
Software Development Intern — API & DevSecOps
- Standardised the company's API landscape organisation-wide — established naming conventions, schema validation, and versioning policies.
- Built Python automation for REST/GraphQL endpoint generation from DB connection strings, saving 40+ hours/week across teams.
- Architected DevSecOps delivery workflow adopted as team standard — integrated SAST, DAST, and dependency scanning into GitLab CI.
- Reduced mean time to deploy by 60% by streamlining multi-environment promotion workflows using ArgoCD.
- Mentored 3 junior interns on clean API design principles, Git workflows, and test-driven development.
Projects
Apr – May 2024
BlinkLink
High-Performance URL Shortening Platform
Production-grade URL shortener with QR codes and real-time analytics. Reduced API latency 96% via Redis caching. Handles 10k+ concurrent requests with sub-10ms p99 latency.
May – Jun 2023
Lucy
AI Voice Assistant
Voice-activated AI assistant with wake-word detection, OpenAI NLU, and OS-level desktop automation. Can browse the web, control applications, and execute system commands via natural language.
Tech Stack
GitHub Activity
What I’m thinking about
Writing on backend engineering, API design, and developer platforms.
Harness Engineering: The Real AI Advantage
The real differentiator in 2026 isn't the model — it's the agent harness around it. Context management, tool orchestration, retry logic, and sub-agent coordination are what make AI reliable in production.
AI Overuse: When Deterministic Logic Trumps GenAI
Not everything needs Generative AI. Knowing when a simple algorithm outperforms an LLM — and having the discipline to choose it — is the real engineering skill in 2026.
LLM Context Rot: Why Attention Fades with Long Chats
LLMs degrade as conversations grow — context rot. The attention mechanism has hard limits, and how you engineer what enters the context window determines the quality of your AI system.
Prompt Engineering vs Context Engineering
They are NOT the same. Prompt engineering is how you ask; context engineering is what you give the model to reason over. Confusing them leads to brittle, unpredictable AI applications.
Dockerfile Best Practices for Performance & Security
Lessons from real CI/CD pipelines: choosing minimal base images, leveraging layer caching, multi-stage builds, and running as non-root can cut image size and attack surface significantly.
MCPs Explained: Why Too Many Tools Confuse AI Agents
MCPs are functions for AI agents — each exposing a name, description, and parameters the LLM reads before deciding what to call. More tools means more decision overhead and more failure modes.
AI Agent Performance Degrades Beyond 128 Tools
Enabling 132 MCPs in Copilot/Cursor triggers a real warning at 128. Every tool adds to the LLM's selection overhead — keep your agent toolset lean, scoped, and purposeful.
AI in Engineering: Balancing Trust and Control
Agentic AI can write your entire codebase while you grab coffee. The power is real — but so is the risk. Knowing when to let the agent run and when to review is the engineering discipline that matters now.
Achievements
Highlights
Smart India Hackathon (SIH) Finalist
2023Competed against thousands of teams nationwide in India's largest hackathon.
Certifications
Certifications in progress — check back soon.
Education
B.Tech Computer Science & Engineering
Contact
Available for Backend & API SDE-1 / SDE-2 roles. Let’s talk.
Whether you have a role in mind, want to collaborate, or just want to talk shop — my inbox is open.
Download Resume