Software Developer, AI/LLM Engineer

Tazwar Sikder

Toronto, ON +1-437-***-**** linkedin.com/in/tazwar89 github.com/tazwar89 ******.******@*******.*** EDUCATION

York University Sep 2023 – Dec 2026

Bachelor of Science in Computer Science (Honours) Toronto, ON TECHNICAL SKILLS

Programming Languages: Python, Java, TypeScript, JavaScript, SQL AI, ML & LLM Systems: RAG pipelines, agentic systems, LangChain, LangGraph, LLM orchestration, NLP, PyTorch, NumPy, Pandas, scikit-learn, Claude, Ollama, vector embeddings, vector search Frameworks & Tools: React, Node.js, FastAPI, Spring Boot, Apache Kafka, Git, JUnit, CI/CD (GitHub Actions) Cloud & Infrastructure: AWS (Elastic Beanstalk, IAM, EC2), Docker, Kubernetes, microservices, Render, Vercel Databases: MySQL, PostgreSQL, MongoDB, SQLAlchemy, Redis (NoSQL), Qdrant (vector database) EXPERIENCE

AI Orchestration Engineer Intern May 2026 – Aug 2026 KRAFZEN Wyoming, USA (Remote)

• Built FastAPI reliability layers and schema validation for LLM pipelines, eliminating crashes from malformed AI outputs.

• Architected an async export orchestration layer for a voice-to-CAD SaaS, raising test coverage from 27% to 66%.

• Fixed critical export status bugs, resolved event schema mismatches, and patched a major JWT security vulnerability.

• Developed a compliance scoring engine with REST endpoints, achieving 85–100% test coverage with automated cleanup.

• Designed an event-driven activity feed across 6 event namespaces and integrated Stripe webhooks for automated billing. Software Developer Intern Jan 2026 – Apr 2026

EduTutor Toronto, ON (Remote)

• Engineered a Dify Ollama fallback inference chain for AI content pipelines, eliminating external paid API costs.

• Built end-to-end RAG pipeline utilizing prompted Ollama LLMs, executing vector evaluations to verify retrieval accuracy.

• Implemented fuzzy near-duplicate question detection across a 500-question window, ensuring quality at scale.

• Resolved 6 critical infrastructure issues across a React/FastAPI stack, eliminating startup crashes and restoring API reliability. Software Developer Intern Sep 2025 – Nov 2025

AllGoodAccounting Brampton, ON

• Owned zero-to-one E2E development of multi-tenant payroll platform, enabling client onboarding for 3 corporate firms.

• Automated core payroll workflows, reducing processing time by 80% and eliminating manual calculation errors.

• Optimized aggregation-heavy MongoDB endpoints, reducing server response times from 1.2s to <200ms. PROJECTS

LogSage Aug 2026 – Present

• Architected a scalable log platform using Kubernetes and Kafka, containerizing 4 microservices with Docker via CI/CD.

• Built a multi-step LangChain/LangGraph agentic system using NLP entity extraction and RAG over a Qdrant vector database.

• Streamed real-time log ingestion via Drain3 template mining, structuring raw text data into Redis-backed JSON.

• Engineered a PyTorch anomaly detector using scikit-learn, NumPy, and Pandas, cutting GenAI call volume by 93%.

• Developed an LLM-as-judge evaluation harness, achieving a 100% pass rate on root-cause diagnosis correctness. SoundWrapped Aug 2025 – May 2026

• Developed a live full-stack SoundCloud analytics platform tracking user listening metrics, genre trends, and artist data.

• Built an LLM data pipeline using Llama models via Groq to aggregate sources and generate artist descriptions.

• Designed a Spring Boot backend service layer integrating 8 external APIs with automated OAuth2 token management.

• Reduced API consumption by 70–90% using multi-layer in-memory caching to bypass platform rate limits.

• Implemented a LastFM ingestion pipeline with 15-minute sync intervals, fuzzy matching, and Supabase persistence.

Contact this candidate