// STATIC VIEW — JAVASCRIPT DISABLED OR UNAVAILABLE
Md Tanvir Hasan Turja
> Machine Learning & Applied AI Engineer · Data Scientist · MLOps
High Wycombe, UK · tanvirh442@gmail.com
// ABOUT
Machine Learning and Applied AI Engineer with 5 years of experience building production
data pipelines, full-stack AI applications, and scalable backend systems. Published author of a
peer-reviewed ACM conference paper, an Open Access journal article in Biomedical Signal Processing
and Control (Elsevier), an arXiv preprint on AMR trend forecasting, a Zenodo-published study on
automotive sales forecasting, and an SSRN preprint proposing an audit battery for LLM-generated
explanations of wearable biosignal alerts.
Founder of Mind Hustle, a live multi-tenant SaaS platform integrating
local vector embeddings over a 226,872-question bank, automated LLM validation pipelines, and Redis job
queues. Holds an MSc in Data Science (Merit, Middlesex University London) with research
experience spanning clinical predictive analytics, LLM-based decision support, and large-scale
epidemiological modelling. Skilled in MLOps, Dockerized infrastructure, and statistical hypothesis
testing across healthcare and commercial domains.
// EXPERIENCE
Founder & Full-Stack Engineer
Mind Hustle (mindhustle.net) · High Wycombe, UK · Dec 2024 – Present
- Architected a multi-tenant SaaS platform serving paying subscribers: Node.js/Express REST API
(50+ route groups, 17 custom middlewares), 40+ MongoDB schemas, Vue 3 SPA and Capacitor mobile apps.
- Integrated local vector similarity search over 226k+ questions using locally hosted transformer
embeddings, LLM-powered question generation, and custom MCP agent tooling that validated
7,145+ questions across 77 batches.
- Built the analytics stack (cohort retention, DAU/MAU, B2B item-analysis reporting), real-time
infrastructure (Redis, BullMQ, Socket.io challenge lobbies) and Stripe billing across 4 pricing
tiers with middleware capability gating.
- Deployed on Hetzner Cloud via Docker Compose and Nginx with automated SSL, maintaining 99%+ uptime.
Tech Lead — Data Infrastructure
Tracer Electrocom (Rainier Technologies) · Bangladesh · Dec 2022 – Aug 2023
- Led data infrastructure evaluation across 4+ concurrent projects; KPI dashboards reduced manual
reporting effort by 30% across 2 stakeholder groups.
- Managed and mentored a team of 5 engineers with zero missed delivery milestones over 12 months.
Full Stack Developer & ML Engineer
Tracer Electrocom (Rainier Technologies) · Bangladesh · Sep 2020 – Nov 2022
- Architected an ML pipeline for healthcare analytics processing 50,000+ patient records, reducing
manual processing time by 40%.
- Deployed a production healthcare system on DigitalOcean with Docker, automated CI/CD, REST model
serving and 80%+ test coverage — zero critical incidents over 18 months.
Internship Teaching Assistant
American International University Bangladesh · Sep 2019 – Dec 2019
- Developed course materials for database systems and mentored 30+ junior developers in database
design and optimisation.
// EDUCATION
MSc in Data Science — Merit
Middlesex University London · 2023 – 2024
Research: Unsupervised Anomaly Detection in Wearable Foot Sensor Data — published in Biomedical Signal
Processing and Control, Elsevier (2026). 🏆 Top 10 Best Poster, Middlesex MSc.
BSc in Computer Science & Engineering — GPA 3.56/4.0
American International University Bangladesh · 2016 – 2021
// PUBLICATIONS
- Auditing LLM Explanations of Biosignal Alerts —
SSRN preprint · 2026
- Unsupervised Anomaly Detection in Wearable Foot
Sensor Data: A Baseline Feasibility Study Towards Diabetic Foot Ulcer Prevention —
Biomedical Signal Processing and Control, Elsevier · 2026
- Forecasting AMR Trends Using ML on WHO GLASS
Data: A RAG Approach for Policy Decision Support — arXiv preprint · 2026
- A RAG-Augmented Hybrid Forecasting Framework for
Policy-Driven Sales Anomaly Explanation — Zenodo · 2026
- Automated Parking System using Graph
Algorithm — ACM International Conference on Computing Advancements · 2020
// FEATURED PROJECTS
- Mind Hustle — multi-tenant gamified learning SaaS with 226k+ question vector
search, Stripe billing, Redis/BullMQ job queues. mindhustle.net
- Foot Ulcer Prediction — unsupervised anomaly detection on time-series temperature
& pressure data with Isolation Forest interpretability, published in
Elsevier.
- AMR Trend Forecasting & Policy RAG — XGBoost on 5,909 WHO GLASS observations
(test MAE 6.13%, R² 0.885, 85.3% error reduction over baseline) with a hallucination-resistant RAG
pipeline achieving 100% citation faithfulness. arXiv
- Automotive Sales Forecasting & Policy RAG — XGBoost/LSTM on 39,496 model-month
records (18.2% MAPE held-out) with multilingual Sentence-BERT retrieval over 1,780 policy articles.
Zenodo
- LLM Alert Explanation Audit — audit battery for an alert-triggered wearable
biosignal pipeline: unsupervised detection, retrieval over a 206-document corpus, and grounded
LLM generation; exposed intra-patient leakage (0.899 AUC → chance inter-patient) and retrieval
poisoning. SSRN
// SKILLS
- ML & AI: PyTorch, TensorFlow, Keras, XGBoost, LightGBM, scikit-learn, LSTM,
Isolation Forest, LOF, one-class SVM, time-series analysis, anomaly detection, pre-registered
evaluation protocols, SHAP / XAI, MLOps, MLflow
- LLMs & NLP: RAG systems, LLM-as-a-judge, RAG security & poisoning probes,
prompt-leakage controls, Sentence-BERT, hybrid BM25 retrieval, vector similarity search,
ChromaDB, Ollama / OpenRouter, hallucination mitigation, Hugging Face
- Data Engineering & Analytics: Python, R, SQL, Apache Spark / PySpark, ETL/ELT,
PostgreSQL, MongoDB, Redis, BigQuery, Pandas, Tableau, Power BI, SciPy
- Full Stack & DevOps: Vue 3, Node.js, Express, FastAPI, Socket.io, Stripe,
Tailwind CSS, Docker, Nginx, CI/CD, Azure, Hetzner Cloud, Firebase
- Domains: healthcare data, wearable sensors, ECG/PPG biosignal analysis, AMR
surveillance, clinical decision support, EdTech, SaaS
// CERTIFICATION
Microsoft Azure Data Scientist Associate (Exam DP-100) — in progress,
expected 2026.
// CONTACT