Builds and experiments
Projects
Hardware, software and models. Some are finished, some are still on the desk.
Hardware
- SesameVoice-controlled robot · personal buildSesame is an open-source conversational robot designed by Dorian Todd. I rebuilt it on my own electronics instead of the original PCB, then wrote the firmware that ties everything together.Built
- Edge AI and roboticsOn-device perception · personal projectReal-time object detection running on-device, where compute is tight. An ESP32 turns the model's output into physical action, so the robot sees something and reacts to it.Built
- Mars Surface Survey UAVAutonomous drone · ISRO Robotics HackathonA drone built for an ISRO-organized hackathon that simulated autonomous surveying of the Martian surface. I worked on software and electronics subsystems and on integrating them on the aerial platform under competition constraints.Competition
AI / ML
- GPT-2 transformerPyTorch · open sourceA mini GPT-2 written from the ground up: custom tokenizer, token and positional embeddings, causal multi-head self-attention, residual feed-forward layers and autoregressive generation. Trained on Tiny Shakespeare.Open source
- BPE tokenizerPythonThe cl100k_base tokenizer used by GPT-4, reimplemented from scratch: subword tokenization, merge rules and vocabulary construction.Built
- Word2VecPythonWord embeddings built without a framework, capturing semantic relationships as the base for similarity and analogy tasks.Built
- BrainCellsPyTorch tensors onlyA neural network built using only raw PyTorch tensors: no torch.nn, no autograd. Forward pass, activations, loss, backprop and weight updates all written by hand.Open source
- Naive Bayes spam classifierNumPy, SciPy, PandasA spam classifier from scratch using word frequencies and Bayes’ theorem, with no ML libraries.Built
- From-scratch implementationsNumPy · ongoingPerceptron, softmax regression, Gaussian discriminant analysis and SGD, written with only NumPy and the math, a repo I keep adding to.Open source