Ramika De Silva
Exploring the intersection of business and technology to build impactful software
I'm currently building my500, a full‑stack language learning platform built with Next.js (App Router), React, TypeScript, PostgreSQL, and Prisma, integrating OpenAI GPT‑4, embeddings, and pgvector for personalized, story‑driven learning.
I focus on building server‑rendered applications, RESTful APIs, relational data models, and production AI workflows, with attention to performance, reliability, and scalability.
Vancouver, BC · UBC
Experience
Roles and internships across software engineering, applied AI, and go-to-market systems.
Huawei
AI Software Engineer Intern
Sept 2026–Dec 2026 · Waterloo, ON
Python, TypeScript, multi-agent systems
- Building an AI-native operator runtime in Python and TypeScript to convert natural language specifications into structured task graphs and executable multi-agent workflows.
- Developing multi-agent orchestration systems coordinating planner, builder, reviewer, and research agents across heterogeneous execution surfaces and model fleets.
- Engineering evidence-gated verification pipelines, benchmark reproduction suites, and execution trace monitors to validate AI-generated artifacts for correctness and reproducibility.
nwPlus
GTM Engineer
May 2025–Present · Vancouver, BC
- Engaged 1,000+ prospective partners and contributed to a $10K+ sponsorship pipeline by architecting automated GTM data pipelines and outbound sequences using n8n and Apollo.io.
Synexus Labs
Software Engineer Intern
Oct 2025–Feb 2026 · San Francisco, California
LangChain, prompt templating, Python
- Led the transition from OpenAI SDKs to provider-agnostic frameworks, using a Dependency Inversion Design Pattern, improving performance and latency across multiple LLM providers with Python.
- Improved benchmark outputs 20% by implementing LangChain prompt templates, chains, and in-memory conversation memory to standardize debugging workflows and reduce response variability across repeated analyses.
- Authored comprehensive technical design documents for LangChain agentic architectures using object-oriented principles to align stakeholders and guide production-ready deployments.
Technical skills
Languages
AI & Machine Learning
Frameworks & Libraries
Tools & Infrastructure
Proof of Work
From idea to shipped builds with measurable progress and learning
Aegis Agent
Built at Cursor Hackathon '26
Autonomous red-team AI agent that analyzes GitHub pull requests for database vulnerabilities, then posts remediation summaries on the PR.
LangGraph state machine driving sqlmap and nuclei scanners via async tool pipelines
About
I've been building software since I was 13, starting with small side projects and gradually moving toward larger, production‑level applications. I'm currently a third‑year Business and Computer Science (BUCS) student at UBC, where my work sits at the intersection of software engineering, systems design, and applied AI.
Most of my experience comes from building full‑stack web applications and AI‑enabled systems end to end. I work primarily with React, TypeScript, Next.js (App Router), PostgreSQL, and Prisma, and I've built backend APIs and data pipelines using Node.js, FastAPI, and Java‑based systems. On the AI side, I've integrated LLMs, embeddings, and vector databases to support personalization, semantic search, and content generation in real user‑facing products.
I'm especially interested in how modern web architecture, relational data models, and applied AI techniques come together in production systems. I enjoy working close to the codebase, iterating on real features, and improving reliability, performance, and scalability over time.
Outside of work, I enjoy learning new languages 💬 and playing electric guitar 🎸
Ideas That Shaped My Work
Books that changed how I think about building products, managing attention, and creating leverage.
Product & Strategy
Principles of Building AI Agents
Sam Bhagwat
Designing AI systems as modular, goal-driven agents rather than static features.
The Mom Test
Rob Fitzpatrick
Good products come from good questions and honest user conversations.
Focus & Mental Models
Deep Work
Cal Newport
Sustained focus is a competitive advantage in a distracted world.
Atomic Habits
James Clear
Small systems, repeated consistently, compound into meaningful change.
What Stuck With Me
Moonwalking with Einstein
Joshua Foer
A look into memory, learning, and how much skill is shaped by training and systems—told through a compelling personal journey.
Writing
Notes on building, learning, and figuring things out as I go.