Data Scientist / Applied ML Engineer ArcLogiq Software's Pvt Ltd
Core
Build and optimize DigiCatalog's probabilistic AI stack by selecting, orchestrating, and optimizing open-source AI models for OCR, multilingual NLP, computer vision, and multimodal pipelines.
Role type
Applied ML Engineer (Multimodal AI & OCR)
Builds
Scalable AI systems for document processing and image attribute extraction
Domain
Computer Vision, NLP, Multimodal AI, Document Intelligence
Deliverable
production ML models
Required skills
Open-source model orchestration, OCR pipeline development, Image attribute extraction, Multilingual AI workflows, Automated evaluation pipelines, Inference optimization, Python, ML engineering
Preferred skills
Edge AI deployments, Vector search and embeddings, GPU optimization, Distributed inference pipelines
Technologies
DocTR, PaddleOCR, TrOCR, Grounding DINO, OWL-ViT, CLIP, Small VLMs, Real-ESRGAN, SDXL-Turbo, IndicTrans2, NLLB, IndicWhisper
Responsibilities
Select and orchestrate open-source models based on use-case fit; Build OCR pipelines using DocTR, PaddleOCR, and TrOCR-class models; Develop image attribute extraction systems using Grounding DINO, OWL-ViT, and CLIP; Work on image enhancement and generation using Real-ESRGAN and SDXL-Turbo; Create automated evaluation pipelines for OCR CER/WER, Attribute F1, and hallucination rates; Optimize ML systems across edge and cloud deployments for cost, latency, and quality.
Seniority
Mid-Senior, hands-on IC