Skip to content
View Anjanamb's full-sized avatar
3 followers · 4 following

Achievements

Achievement: YOLOAchievement: QuickdrawAchievement: Pull Sharkx2Achievement: Pair Extraordinaire

Achievements

Achievement: YOLOAchievement: QuickdrawAchievement: Pull Sharkx2Achievement: Pair Extraordinaire

Highlights

  • Pro

Block or report Anjanamb

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. Markdown is supported. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
Anjanamb/README.md

Hi, I'm Anjana 👋

Typing SVG

Portfolio LinkedIn Email Location Profile views

I build intelligent systems that bridge hardware and software. Currently focused on reducing hallucinations in task-oriented dialogue via preference-based optimisation, industrial IoT automation, and applied ML for time-series problems.

🔬 What I'm working on

  • 🧠 Lab Rotation 2 @ HHU — DPO-based hallucination reduction on dialogue NLG. Strict preferences beat gold/random by 4–6× on faithfulness; working paper underway.
  • 🏭 RIF Institut, Dortmund — Containerised brewery process control: PLC over MQTT, time-series replay via FastAPI + InfluxDB, Streamlit ops dashboards.
  • 📈 Recently wrapped — TFT + PPO pairs trader: an end-to-end, CI-tested forecasting + RL stat-arb pipeline whose look-ahead-free walk-forward honestly showed no edge over a SPY buy-and-hold (the value is the rigour, not a backtest Sharpe).

⭐ Featured work
🖥️ wattson DL-workload-aware system monitor — textual TUI with GPU-aware process table, NVML temps, 1 Hz refresh. Active development.
📈 tft-ppo-pairs-trader Temporal Fusion Transformer spread forecaster + PPO RL agent for multi-asset pairs trading (DuckDB · Gymnasium · Optuna/MLflow · Docker · CI, 41 tests). Built a look-ahead-free walk-forward backtester — hedge ratio and TFT refit per fold — that honestly showed the strategy has no edge over a SPY buy-and-hold: a pipeline rigorous enough to reject its own false positives.
🔧 sensor-anomaly-detection · 🚀 demo Unsupervised anomaly detection on NASA C-MAPSS turbofans. 5 detectors across FD001–FD004; SHAP attribution surfaced that the model picks up HPC degradation through bypass-duct pressure before direct sensors drift. Streamlit + Power BI surfaces.
🔬 Digital-IC-NN-Duplication NN-based digital twin for sequential circuits. Published at IEEE 2023.
🖼️ image-downsampling-processor Verilog low-pass + downsample-by-2 processor, validated against Python reference via SSD.

🛠️ Tech I reach for

ML / Deep Learning

PyTorch HuggingFace scikit-learn TensorFlow spaCy OpenCV

Languages

Python C++ C# JavaScript Verilog MATLAB

Data & Infrastructure

FastAPI Docker InfluxDB DuckDB PostgreSQL MySQL Streamlit

IoT · Embedded · Networking

Node-RED MQTT Raspberry Pi Cisco

Cloud · Tools

Azure AWS Git Linux

📊 GitHub stats

GitHub streak

Activity graph

💡 Curious devs: poke around the portfolio — toggle ▮ terminal mode and try help. There are easter eggs.

Pinned Loading

  1. covid-19-realtime-nodered-dashboard covid-19-realtime-nodered-dashboard Public

    Real-time COVID-19 dashboard using IoT primitives — Node-RED + MQTT + ESP8266 + REST APIs.

    C++ 1

  2. Digital-IC-Functionality-Duplication-Using-NN Digital-IC-Functionality-Duplication-Using-NN Public

    Reproduce sequential digital circuits with neural networks trained on I/O traces — a functional digital twin. IEEE 2023.

    Python 1

  3. custom-image-downsampling-processor custom-image-downsampling-processor Public

    Verilog custom processor for low-pass filtering + downsampling-by-2 of images. Verified against Python reference via SSD.

    Verilog

  4. FPGA-Implementation-and-Interfacing FPGA-Implementation-and-Interfacing Public

    Collection of FPGA designs, testbenches, and interfacing experiments validating hardware implementations of digital circuits.

    Verilog 1

  5. Flask-Image-Classifier Flask-Image-Classifier Public

    Flask web app for real-time image classification using pretrained MobileNetV2 (ImageNet weights).

    Python

  6. wattson wattson Public

    Your machine's personal assistant — DL-workload-aware system monitor (TUI)

    Python