Aditya Kulkarni
AI Engineer · Speech and Language
Professional Summary
AI engineer with 4 years building speech and language systems for Indian languages. Trains and serves speech-recognition and translation models for call centres and voice apps.
Work Experience
AI Engineer, Vaanika Voice Labs
- Fine-tuned a Whisper model on 3,000 hours of Telugu and Hindi call audio, cutting word error rate from 28% to 14%.
- Built a streaming speech-to-text service on Triton serving 1,800 concurrent calls with under 400ms delay.
- Added an LLM call-summary step that saved agents about 90 seconds per call.
Machine Learning Engineer, Godavari Datalabs
- Trained intent classifiers for a banking voice bot in 5 languages, reaching 91% intent accuracy.
- Built the data-cleaning pipeline that removed 22% noisy transcripts before training.
Education
| Institution | Degree | Dates | Score/GPA |
|---|---|---|---|
Deccan Plateau Institute of Technology | B.Tech, Computer Science and Engineering | 2016 – 2020 | CGPA 8.5/10 |
Skills
Speech & NLP
Whisper·wav2vec 2.0·NeMo·Hugging Face
Serving
Triton·ONNX Runtime·Kubernetes·gRPC
Languages
Python·C++·SQL
Data labelling guidelines·Mentoring
Publications
- Low-resource code-switched ASR for Telugu–English call audio — Regional Workshop on Speech Technology, 2024
Download your resume to print it.
Chip
ATS-safeA clean data-sheet layout: contact details, skills and education in neatly aligned label and table rows, with your experience told as a clear narrative. Single column, ATS-safe.
Single columnTable layout
- No photo — text only.
- Page size: A4.
- Switching to another template later keeps all your content.