DigiTap is Hiring Junior Data Scientist - Computer Vision | Bangalore | Freshers

Job Description

DigiTap's Computer Vision opening stands out for its 𝐝𝐨𝐦𝐚𝐢𝐧 𝐬𝐩𝐞𝐜𝐢𝐟𝐢𝐜𝐢𝐭𝐲 — this isn't generic "CV for images" work, it's applied to 𝐟𝐢𝐧𝐭𝐞𝐜𝐡 𝐚𝐧𝐝 𝐝𝐨𝐜𝐮𝐦𝐞𝐧𝐭 𝐢𝐧𝐭𝐞𝐥𝐥𝐢𝐠𝐞𝐧𝐜𝐞, meaning OCR, document understanding, and image quality assessment are core focuses rather than side skills. For freshers deciding between multiple CV/ML openings, this matters: document-AI and fintech-specific computer vision is a narrower, more specialized niche than general object detection or classification work, and experience here doesn't transfer as directly to, say, autonomous vehicles or medical imaging roles — but it's a genuinely in-demand skill set given how many fintech and RegTech companies now need document verification and OCR pipelines.

The role is structured across the 𝐟𝐮𝐥𝐥 𝐌𝐋 𝐥𝐢𝐟𝐞𝐜𝐲𝐜𝐥𝐞 — from data preparation through model training, evaluation, and production deployment support — rather than being narrowly research-focused or narrowly engineering-focused. For most freshers coming from academic CV projects, the biggest gap to close is likely the 𝐩𝐫𝐨𝐝𝐮𝐜𝐭𝐢𝐨𝐧/𝐝𝐞𝐩𝐥𝐨𝐲𝐦𝐞𝐧𝐭 𝐬𝐢𝐝𝐞 (API integration, model inference pipelines, performance optimization for production) since most college projects stop at training and evaluating a model in a notebook, without ever taking it to a deployed, monitored system.

𝐒𝐤𝐢𝐥𝐥𝐬-𝐰𝐢𝐬𝐞, the listing draws a useful distinction between "hands-on experience" (Python, PyTorch) and "basic understanding" (CNNs, YOLO, Faster R-CNN, transfer learning) — meaning DigiTap isn't expecting you to have already deployed production object detection systems, but does expect genuine conceptual grounding beyond just having taken an online course. Given how many candidates now claim "computer vision experience" from a single Coursera certificate, a specific academic project or internship involving actual OCR or document processing (explicitly called out as a preferred qualification) is likely to meaningfully differentiate an application in this pool.

𝐎𝐮𝐫 𝐓𝐚𝐤𝐞:

- 𝐆𝐨𝐨𝐝 𝐟𝐨𝐫: CS/AI graduates with genuine PyTorch and OpenCV project experience who are specifically interested in document intelligence/OCR rather than pure computer vision research

- 𝐍𝐨𝐭 𝐢𝐝𝐞𝐚𝐥 𝐟𝐨𝐫: those with only theoretical ML coursework and no hands-on framework experience — the role explicitly wants demonstrated Python/PyTorch proficiency, not just conceptual familiarity

- 𝐒𝐞𝐥𝐞𝐜𝐭𝐢𝐨𝐧 𝐝𝐢𝐟𝐟𝐢𝐜𝐮𝐥𝐭𝐲: moderate — the technical bar (PyTorch, OpenCV, CNN fundamentals) is achievable for a dedicated CV/ML fresher, but the fintech/document-processing specialization narrows the pool of candidates with directly relevant project experience

- 𝐂𝐚𝐫𝐞𝐞𝐫 𝐭𝐫𝐚𝐣𝐞𝐜𝐭𝐨𝐫𝐲: commonly leads into Computer Vision Engineer, ML Engineer, or Applied Scientist roles, with document intelligence/OCR specialization being a valuable niche given growing fintech and RegTech demand for automated document verification

𝐂𝐨𝐦𝐩𝐞𝐧𝐬𝐚𝐭𝐢𝐨𝐧 𝐂𝐨𝐧𝐭𝐞𝐱𝐭: DigiTap hasn't disclosed pay for this role. Junior Data Scientist/Computer Vision roles at fintech startups in Bangalore typically range around ₹6–10 LPA for candidates with 0-2 years of experience and demonstrable project work, based on general market data for comparable ML/CV fresher roles — this is an estimate only and can vary significantly based on the strength of your project portfolio and DigiTap's specific compensation structure as a smaller company relative to larger tech firms.

𝐀𝐩𝐩𝐥𝐢𝐜𝐚𝐭𝐢𝐨𝐧 𝐒𝐭𝐫𝐚𝐭𝐞𝐠𝐲: Since OCR/document processing exposure is explicitly called out as preferred, if you have any academic or personal project involving document analysis, text extraction, or even a Kaggle competition using OCR datasets, lead with that specifically rather than generic "I did an image classification project" claims — given how specialized DigiTap's actual use case is (fintech document intelligence), demonstrating any prior exposure to this specific problem space is likely to carry more weight than broader CV experience alone.
The role is structured across the full ML lifecycle — from data preparation through model training, evaluation, and production deployment support — rather than being narrowly research-focused or narrowly engineering-focused. For most freshers coming from academic CV projects, the biggest gap to close is likely the production/deployment side (API integration, model inference pipelines, performance optimization for production) since most college projects stop at training and evaluating a model in a notebook, without ever taking it to a deployed, monitored system.

Skills-wise, the listing draws a useful distinction between "hands-on experience" (Python, PyTorch) and "basic understanding" (CNNs, YOLO, Faster R-CNN, transfer learning) — meaning DigiTap isn't expecting you to have already deployed production object detection systems, but does expect genuine conceptual grounding beyond just having taken an online course. Given how many candidates now claim "computer vision experience" from a single Coursera certificate, a specific academic project or internship involving actual OCR or document processing (explicitly called out as a preferred qualification) is likely to meaningfully differentiate an application in this pool.

Our Take:

Good for: CS/AI graduates with genuine PyTorch and OpenCV project experience who are specifically interested in document intelligence/OCR rather than pure computer vision research
Not ideal for: those with only theoretical ML coursework and no hands-on framework experience — the role explicitly wants demonstrated Python/PyTorch proficiency, not just conceptual familiarity
Selection difficulty: moderate — the technical bar (PyTorch, OpenCV, CNN fundamentals) is achievable for a dedicated CV/ML fresher, but the fintech/document-processing specialization narrows the pool of candidates with directly relevant project experience
Career trajectory: commonly leads into Computer Vision Engineer, ML Engineer, or Applied Scientist roles, with document intelligence/OCR specialization being a valuable niche given growing fintech and RegTech demand for automated document verification
Compensation Context: DigiTap hasn't disclosed pay for this role. Junior Data Scientist/Computer Vision roles at fintech startups in Bangalore typically range around ₹6–10 LPA for candidates with 0-2 years of experience and demonstrable project work, based on general market data for comparable ML/CV fresher roles — this is an estimate only and can vary significantly based on the strength of your project portfolio and DigiTap's specific compensation structure as a smaller company relative to larger tech firms.

Application Strategy: Since OCR/document processing exposure is explicitly called out as preferred, if you have any academic or personal project involving document analysis, text extraction, or even a Kaggle competition using OCR datasets, lead with that specifically rather than generic "I did an image classification project" claims — given how specialized DigiTap's actual use case is (fintech document intelligence), demonstrating any prior exposure to this specific problem space is likely to carry more weight than broader CV experience alone.

Roles & Responsibilities

The responsibilities span the complete model lifecycle rather than a single narrow function — you'd be involved from raw data preparation through to production monitoring, which is broader hands-on exposure than many junior CV roles offer.

- Build and improve computer vision models for object detection, OCR, document understanding, and image classification
Collect, clean, label, and analyze image/document datasets, identifying data quality issues

- Train, fine-tune, and evaluate deep learning models, benchmarking different approaches

-Read and implement research papers, experimenting with new architectures

-Collaborate with engineering teams on model deployment, monitoring, and API integration

-Document experiments and findings while working closely with product and data science teams

Qualifications & Eligibility

The degree requirement is specifically technical (no business/generalist path here, unlike some AI-adjacent roles), reflecting the hands-on technical nature of the position.

-BE/BTech/ME/MTech/MCA in Computer Science, AI, or related fields

-0–2 years of experience in Computer Vision, AI, or Machine Learning (preferred, not strictly mandatory)

-Internship or academic projects in Deep Learning or Computer Vision (preferred)

-Exposure to OCR, document processing, or image analytics projects (preferred, and a genuine differentiator)

Skills Required

PyTorch and core CV libraries form the practical toolkit expected; model architecture knowledge is expected at a foundational rather than expert level.

- Strong ML/Deep Learning fundamentals and Computer Vision concepts
Hands-on Python programming; PyTorch preferred among deep learning frameworks

-CV libraries: OpenCV, NumPy, Pandas, PIL, scikit-image, matplotlib
Basic understanding of CNNs, object detection (YOLO, Faster R-CNN, SSD), and transfer learning

-Strong analytical and debugging skills