Multilingual Speech AI Engineer
Budget: ₹12,500 – ₹37,500 INR
# Speech AI / Multilingual Cognitive AI Engineer (Contract – 1 Month)
## About Us
We are building an AI-powered multilingual cognitive decline screening platform that analyzes short speech recordings to identify early cognitive impairment using speech and language biomarkers.
Our current platform already includes:
* FastAPI backend
* Whisper-based ASR pipeline
* Acoustic and linguistic feature extraction
* Cognitive scoring pipeline
* Web interface
* RunPod deployment
* Firebase integration
* English MVP
We are now looking for a Speech AI engineer to improve the core multilingual intelligence of the platform.
---
## Responsibilities
* Evaluate the existing ASR and cognitive screening pipeline
* Improve transcription robustness for English, Hindi, and Hinglish
* Improve handling of code-switched speech
* Optimize Whisper or evaluate alternative multilingual ASR approaches
* Build confidence scoring and error handling for low-confidence transcripts
* Improve acoustic and linguistic feature extraction
* Benchmark model performance using objective evaluation metrics
* Document findings and recommendations
* Work closely with the founding team on iterative experiments
---
## Required Skills
* Python
* PyTorch
* Speech AI / ASR
* Whisper or similar speech models
* Hugging Face
* Audio processing (Librosa, Torchaudio, etc.)
* FastAPI
* Git
---
## Preferred
* Experience with multilingual ASR
* Hindi/Hinglish speech processing
* Code-switching research
* Healthcare AI or cognitive assessment
* Model evaluation and benchmarking
---
## Deliverables
By the end of the engagement, the engineer should deliver:
* Improved multilingual ASR pipeline
* Better English, Hindi, and Hinglish performance
* Benchmark report with accuracy metrics and failure analysis
* Production-ready inference improvements
* Fully documented code integrated into the existing VoiceMind repository
---
## Duration
1 Month (Remote)
## Compensation
₹15,000 (Fixed Contract)
Outstanding performance may lead to a longer-term research collaboration.
## About Us
We are building an AI-powered multilingual cognitive decline screening platform that analyzes short speech recordings to identify early cognitive impairment using speech and language biomarkers.
Our current platform already includes:
* FastAPI backend
* Whisper-based ASR pipeline
* Acoustic and linguistic feature extraction
* Cognitive scoring pipeline
* Web interface
* RunPod deployment
* Firebase integration
* English MVP
We are now looking for a Speech AI engineer to improve the core multilingual intelligence of the platform.
---
## Responsibilities
* Evaluate the existing ASR and cognitive screening pipeline
* Improve transcription robustness for English, Hindi, and Hinglish
* Improve handling of code-switched speech
* Optimize Whisper or evaluate alternative multilingual ASR approaches
* Build confidence scoring and error handling for low-confidence transcripts
* Improve acoustic and linguistic feature extraction
* Benchmark model performance using objective evaluation metrics
* Document findings and recommendations
* Work closely with the founding team on iterative experiments
---
## Required Skills
* Python
* PyTorch
* Speech AI / ASR
* Whisper or similar speech models
* Hugging Face
* Audio processing (Librosa, Torchaudio, etc.)
* FastAPI
* Git
---
## Preferred
* Experience with multilingual ASR
* Hindi/Hinglish speech processing
* Code-switching research
* Healthcare AI or cognitive assessment
* Model evaluation and benchmarking
---
## Deliverables
By the end of the engagement, the engineer should deliver:
* Improved multilingual ASR pipeline
* Better English, Hindi, and Hinglish performance
* Benchmark report with accuracy metrics and failure analysis
* Production-ready inference improvements
* Fully documented code integrated into the existing VoiceMind repository
---
## Duration
1 Month (Remote)
## Compensation
₹15,000 (Fixed Contract)
Outstanding performance may lead to a longer-term research collaboration.