ECE Final-Year @ BMS College of Engineering, Bengaluru
π― Focused on AI-driven Signal Processing, Communication Systems & Intelligent DSP
Iβm fascinated by how information travels β through air, wires, or neurons β and how intelligence can adapt to that noise.
My work blends machine learning, signal processing, and system design, building end-to-end frameworks that learn from and react to dynamic signals.
Iβm currently aligning my research and engineering work toward and aiming to specialize in AI for Communication & Signal Systems.
| Project | Domain | Current Focus |
|---|---|---|
| π° SkyTune | AI for SATCOM | Learning-based ModCod adaptation under fading conditions |
| π§© SigFlow | Software Signal Pipeline (Python + C++20) | Integrating DSP β ML for adaptive signal chains |
| π§ Neural Transceiver | Deep Comms Learning | Autoencoder-based encoder-decoder over AWGN/Rayleigh channels |
| π Neural Channel Estimation | OFDM + CNN | CNNs for channel-tap inference & BER optimization |
| π‘ Anomaly Detection in Sensor Streams | Time-series ML | LSTM Autoencoder for industrial signal anomalies |
- Signal Intelligence: adaptive filtering, FFT/OFDM, Rayleigh/AWGN modeling
- Learning Systems: CNNs, LSTMs, Transformers, Autoencoders, RL agents
- Integration: Python β C++20 (PyBind11), MATLAB/CommPy simulations
- Applied Math: Probability, Linear Algebra, Optimization, Information Theory
Core Libraries: NumPy Β· SciPy Β· scikit-learn Β· FFTW Β· Eigen Β· PyBind11 Β· TensorBoard
- βοΈ SigFlow reduced DSP latency by ~30 % using PyBind11 C++20 core
- π° SkyTune prototype improved throughput by ~18 % at equal BER
- π§ EEG Classification achieved 92 % F1-score using spectral features
- π Adaptive LMS Canceller delivered ~20 dB SNR gain on live audio
- π Top 5 % nationwide β NPTEL Machine Learning (IIT-Madras)
- AI-based Communication Systems
- Neural Channel Estimation & Modulation Recognition
- Adaptive DSP and RL-driven PHY optimization
- Biomedical & Acoustic Signal Intelligence
| Phase | Timeline | Focus |
|---|---|---|
| Phase 1 β Integration | Nov β Dec 2025 | Build & benchmark SigFlow v1.0 (Python + C++20) |
| Phase 2 β Domain Depth | Jan β Feb 2026 | Paper + Experiment sprint on AI-Comms / Signal AI |
| -- Rest will be updated later |
- πΈ Amateur music producer β experiments with synth-based sound design
- π² Love a good long ride
- π An avid swimmer
βBuild systems that understand their own noise.β
β Dibyojyoti Bhattacharjee
