Kyndryl
Data Scientist / Applied Machine Learning Engineer
The role involves developing AI pipelines for OCR, image understanding, image enhancement, multilingual language processing, and Vision-Language Models (VLMs) using modern open-source technologies. The candidate will also establish automated model evaluation frameworks, perform rigorous benchmarking, and continuously improve model quality through error analysis, calibration, and optimization techniques. Key Responsibilities Design, develop, and deploy end-to-end Machine Learning, Computer Vision, NLP, and Multimodal AI solutions using open-source models. Build AI pipelines for OCR, image attribute extraction, image enhancement, image generation, and multilingual language processing. Develop automated evaluation frameworks using metrics such as OCR CER/WER, Attribute F1 Score, Hallucination Rate, and regression testing. Optimize AI models for accuracy, scalability, cost, and latency across cloud and edge environments using threshold tuning, calibration, ensembling, and post-processing techniques. Collaborate with product managers, data engineers, and cross-functional teams to evaluate emerging AI technologies and deliver production-ready AI solutions. Mandatory Requirements 3–6 years of hands-on experience in Applied Machine Learning, Computer Vision, NLP, or Multimodal AI. Strong proficiency in Python, PyTorch/TensorFlow, and modern ML frameworks. Hands-on experience with OCR technologies (DocTR, PaddleOCR, TrOCR), Vision-Language Models (Grounding DINO, OWL-ViT, CLIP), and multilingual AI models such as IndicTrans2, NLLB, or IndicWhisper. Strong expertise in model evaluation, benchmarking, error analysis, regression testing, model optimization, and performance tuning. Experience with Git, Docker, REST APIs, and deploying AI/ML solutions in production environments.
The role
What you'll do
- Design, develop, and deploy end-to-end Applied ML, Computer Vision, NLP, and Multimodal AI solutions using open-source models.
- Build and orchestrate AI pipelines for OCR, image attribute extraction, image enhancement, generation, and multilingual language processing.
- Develop and maintain automated evaluation frameworks, including OCR CER/WER, Attribute F1, Hallucination Rate, regression testing, and model benchmarking.
- Optimize AI models and pipelines for cost, latency, scalability, and accuracy across edge and cloud environments using threshold tuning, calibration, ensembling, and post-processing techniques.
- Collaborate with product managers, data engineers, and business stakeholders to research, evaluate, and implement the latest open-source AI models and ensure successful production deployment.
You bring
Requirements
- 3–6 years of hands-on experience in Applied Machine Learning, Computer Vision, NLP, or Multimodal AI.
- Strong proficiency in Python, PyTorch/TensorFlow, and modern ML frameworks with experience in deploying production-grade AI solutions.
- Hands-on experience with OCR technologies (DocTR, PaddleOCR, TrOCR), Vision-Language Models (CLIP, Grounding DINO, OWL-ViT), and multilingual AI models such as IndicTrans2, NLLB, or IndicWhisper.
- Strong expertise in model evaluation, error analysis, benchmarking, regression testing, threshold tuning, calibration, and performance optimization.
- Experience with Git, Docker, REST APIs, and MLOps/deployment practices, along with the ability to quickly evaluate and integrate new open-source AI models into production.
Skills