Software & Systems Engineer / ML Systems • Full Stack• Low & High Level

Montreal, QC · CV Projects · taha.slime@gmail.com

I build fast, reliable systems end‑to‑end. From C/epoll servers and x86_64 assembly to Python ML training and model deployment.

• Fine‑tuning compact LLMs (Phi‑2,TinyLLama,Bert-Mini) with LoRA + 4‑bit; REST integration
• BPE tokenizer in Python & C++ (max‑heap + custom hash table, benchmarks)
• Vector database from scratch (L2/cosine, simple APIs, RAG hooks)
• Cross‑language socket programming (C/C++ epoll, Java NIO, Python asyncio, Node.js)
• SIMD float calculator in x86_64 (NASM, SSE; manual parsing/formatting)
• Redis‑compatible key‑value store in C (RESP parser, epoll, open‑addressing hash table)

• Fast backend infra (Redis clone, tokenizer engine, vector DB)
• From-scratch implementations (hash tables, epoll servers, RESP parser, BPE)
• End-to-end ML pipelines (LoRA, tokenization, deployment)
• Cross-language low-level systems (C, C++, Python, NASM, Java, Node.js)

• Neural Network from Scratch — Minimal deep learning framework implementing backpropagation, gradient descent, and activation functions using NumPy.
  🔗 Repo: tslime/NeuralNetworkModel

• BPE Tokenizer (Python & C++) — Custom subword tokenizer using Byte Pair Encoding with heap-based pair frequency tracking and hash table indexing.
  🔗 Repo: tslime/BPEAlgorithm

• VectorDB (C++) — Vector database built from scratch using Hierarchical Navigable Small World (HNSW) graphs. Features index-based design, custom heaps, and efficient nearest-neighbor search.
  🔗 Repo: tslime/VectorDB

• SIMD Matrix Algorithm (x86_64) — NASM + SSE/AVX optimized matrix operations with loop unrolling, memory alignment, and RDTSC benchmarking.
  🔗 Repo: tslime/SIMDMatrixAlgorithm

• Redis Clone (C) — Minimalist Redis-compatible server built from scratch with custom TCP networking, RESP parser, and multi-client concurrency.
  🔗 Repo: tslime/vRedisClone

• Socket Programming Analysis — Cross-language networking study (C, C++, Java, Python) using concurrency models such as epoll, fork, NIO, and asyncio.
  🔗 Repo: tslime/NetworkProgramming

• FastCRUDAPI — Node.js and Express-based RESTful API featuring both JSON and MongoDB backends. Includes input validation, Helmet/CORS security, rate limiting, and integrated Prometheus–Grafana monitoring for real-time observability.
  🔗 Repo: tslime/FastCRUDAPI

• Secure Inference Gateway — FastAPI-based secure gateway for serving local AI models (Ollama). Supports API-key authentication, model discovery, centralized logging, and compliance-ready configuration.
  🔗 Repo: tslime/SecureInferenceGateway

• Data Provenance Tracker — Lightweight tool to log, hash, and audit data transformations for integrity and reproducibility. Supports SHA-256, timestamping, and JSONL-based provenance chains.
  🔗 Repo: tslime/Datatracker

• PID Controller Simulator (C++) — Real-time closed-loop feedback system visualized with Gnuplot. Demonstrates proportional, integral, and derivative tuning and system stability.
  🔗 Repo: tslime/PIDControllerSimulator

• Flight Recorder / Telemetry (C++) — Aircraft telemetry simulator with binary packet serialization, CRC-16 validation, and live altitude/velocity plotting via Gnuplot.
  🔗 Repo: tslime/FlightRecorder

• Special Data Structures — Experimental implementations of advanced structures (custom hash tables, max-heaps for BPE, HNSW indexes).
  🔗 Repo: tslime/SpecialDataStructures

• Fundamental Algorithms — Core computer-science algorithms implemented across multiple languages for clarity and benchmarking.
  🔗 Repo: tslime/FundamentalAlgorithms

Languages: C, C++, Python, Java, Rust
Systems Programming: x86_64 NASM (SIMD, SSE), socket programming (epoll, NIO, asyncio), Redis protocol (RESP), compilers/parsers, OS-level tools, embedded & robotics
Machine Learning: Neural networks, ML algorithms (regression, SVM, trees, clustering), deep learning, synthetic data pipelines, LoRA, 4-bit quantization, PyTorch, TensorFlow
Full Stack Web: Node.js, Express, PostgreSQL, MySQL, HTML/CSS, JS/TS (React, Node.js, Express), API REST, PHP, ASP.NET, Spring Boot
Infra & DevOps: MLFlow, Docker, Kubernetes, Terraform, DVC, AWS/Azure, GitHub, CI/CD
Tools: Git, GitHub, LaTeX, CircleCI
OS: Linux, Windows, MacOS