Audio team lead at Abilix (one of the world's top 5 AI educational robot companies): Led the development and implementation of a audio pipeline, including key components such as:* Audio Classification: (detect wake words) * Speech-to-Text (STT): (transcribing query) * Language Modeling: (generating response) * Text-to-Speech (TTS): (synthesise speech) * spoken answer and voice assistantsImplement or improve the following AI tasks1) Different audio tasks2) Text… Show more Audio team lead at Abilix (one of the world's top 5 AI educational robot companies): Led the development and implementation of a audio pipeline, including key components such as:* Audio Classification: (detect wake words) * Speech-to-Text (STT): (transcribing query) * Language Modeling: (generating response) * Text-to-Speech (TTS): (synthesise speech) * spoken answer and voice assistantsImplement or improve the following AI tasks1) Different audio tasks2) Text classification (NLP)3) Question answering (NLP)4) Summarization (NLP)5) Translation (NLP)6) Recommendation systems (NLP)7) Image classification (computer vision)8) Object detection (computer vision)9) Other LLM, OpenAI langchain, langgraph chatbotTechnologies used:waveform, spectrogram, DFT, STFT, resampling, filtering, preprocess, normalization, feature extractor, attention mast, model-specific tokenizers, stream audio data, custom data transformations, data collator, MSE, MAE, MOS evaluation matrix, Pretrain Transformer models, fine-tunepretrained model for specific task, custom algorithms design, custom head designModels used:Whisper, Vocoder, facebook/voxpopuli, Wav2Vec2, BERT, HuBERT, LibriSpeech, Seq2Seq, SpeechT5, M-CTC-T, HiFi-GANDatasets:PolyAI/minds14, google/fleurs, XLSR, LibriTTS, LJSpeech, Multilingual LibriSpeech, VoxPopuli Show less

completed one ML/computer vision (medical image analysis) projectand 10 web applications for Microsoft's clients, utilizing various JavaScript and Java frameworks, Flutter, Angular, React, Python, and Machine Learning, Image processing, Deep learning

Full-stack Architect and Developer within Google's Network Team, tasked with constructing internal applications for network hardware supply chain management. Leveraged Angular and Java technologies for implementation

I spearheaded the development of approximately 30 internal applications for the company, employing Java and JavaScript technologies.

Building Global Risk Management System, using C++ and Java