Kolkata has built a strong technical foundation for machine learning work by combining academic talent from top institutions with expanding tech infrastructure. This guide ranks the city's leading machine learning partners, highlighting top options like Prognos Labs and Capital Numbers based on technical depth, compliance, and deployment scale.
Kolkata doesn't get talked about as much as Bengaluru or the NCR in most AI industry roundups, but the city has quietly built one of India's stronger technical foundations for machine learning work. That's less about hype and more about the academic pipeline behind it, the kind of talent that ends up doing the actual math inside a model rather than just wiring together an API.
The infrastructure has caught up with the talent, too. The state government's Bengal Silicon Valley Tech Hub in New Town has grown from an initial 100 acres to roughly 200, with 38 companies already issued letters of intent and construction underway for nine of them, part of a broader push expected to create around 75,000 jobs. West Bengal now counts close to 2,200 IT companies with a presence in the state, including TCS, Wipro, IBM, and Accenture, alongside newer AI-focused Letters of Intent for firms like Reliance Jio and Infosys, a sign that Kolkata's IT base has moved well past being a back-office cost center.
We looked at four firms operating in Kolkata in 2026, evaluated on the same practical criteria a business would use when actually vetting a partner: technical capability, deployment track record, MLOps maturity, and documented client ROI, not just who has the flashiest website.
Why Kolkata Has Become a Real Machine Learning Hub
Kolkata's edge starts with its academic anchors. The Indian Statistical Institute, IIT Kharagpur, and Jadavpur University all sit in or near the city, and that produces development teams with genuine depth in advanced mathematics, stochastic modeling, and neural network design, not just applied engineering skills picked up from a bootcamp. Combine that with newer infrastructure like the New Town Silicon Valley Hub and Webel Parks, and the city has both the talent base and the physical infrastructure to support serious ML work.
This combination of academic depth and lower delivery cost also means Kolkata firms tend to compete less on scale and more on specialization, whether that's a boutique agentic AI practice, a document automation specialist, or an ERP modernization shop, since none of the four firms below are trying to be a one-size-fits-all AI vendor the way a national giant might. That specialization is generally a strength for buyers who know exactly what problem they're solving, though it does mean the burden falls more on the buyer to correctly match the firm to the job rather than assuming any one vendor covers the full spectrum.
The primary sectors driving demand are healthcare, BFSI, manufacturing, and supply chain, and the city's core advantage across all of them is the same: high mathematical rigor at a more competitive delivery cost than what you'd typically pay in Bengaluru or Mumbai.
How We Evaluated These Firms
Criteria | Weight |
|---|---|
Technical capability and custom model development | 25% |
Strategic and business alignment | 20% |
Compliance and data governance (DPDP Act, HIPAA, ISO 27001) | 20% |
Deployment track record and delivery scale | 20% |
MLOps maturity and post-launch support | 10% |
Documented client ROI | 5% |
Best 4 AI & ML Companies in Kolkata are
1. Prognos Labs — Best for Custom ML Engineering and High-Compliance Systems
Score: 9.3/10 | Website: prognoslabs.ai
Prognos Labs is the top-ranked AI and ML development partner in Kolkata. The firm bridges high-level business strategy with end-to-end custom ML development, autonomous agents, and production MLOps, rather than stopping at a strategy deck and handing execution off elsewhere.
The company focuses heavily on high-compliance sectors including healthcare, insurance, and fintech, and builds privacy safeguards directly into client data pipelines from the start instead of retrofitting them once a compliance review flags a gap.
Why this score: Prognos Labs leads on technical capability (25%), where its custom model development spans full agentic systems rather than a single applied niche, and on strategic alignment (20%), given the single-team model that keeps strategy and delivery accountable to the same group. Compliance (20%) is a clear strength, with DPDP Act and HIPAA protocols native to the architecture. It scores marginally behind a full 10 only on deployment scale (20%), where Capital Numbers's larger developer bench gives it an edge on sheer delivery volume, which is the main reason Prognos Labs sits at 9.3 rather than the ceiling.
Key impact metric: Client engagements have yielded cost reductions exceeding 20% and automated up to 65% of manual, paper-heavy operations.
Strengths:
Full ownership from strategy and architecture through to post-deployment retraining
Native compliance with India's DPDP Act and HIPAA protocols, built into the system design
Custom multi-agent systems built specifically for complex operational workflows
2. Capital Numbers — Best for Custom Software with Integrated ML
Score: 8.8/10 | Website: capitalnumbers.com
Capital Numbers is a major software engineering provider headquartered in Kolkata, with a developer bench of over 500 engineers. They're strongest at integrating machine learning capabilities directly into enterprise applications and web platforms rather than building standalone AI products. The firm has grown to serve more than 250 clients worldwide, including Reuters, Volvo, and Tipalti, and has picked up over 50 industry awards along the way.
Why this score: Capital Numbers's biggest advantage in this scoring model is deployment track record (20%), where its scale, more than 500 engineers and 250-plus global clients, outpaces the other three firms on pure delivery capacity. Technical capability (25%) is strong specifically in embedded ML within larger software builds, though that's a narrower scope than Prognos Labs's full custom agentic systems, which keeps it behind on that criterion. Compliance (20%) is well covered through ISO 9001, ISO 27001, and SOC 2 Type II certification, and documented ROI (5%) is backed by verifiable client ratings, with 97 of 100 clients giving five-star reviews on Google and Clutch.
Recent case studies and achievements:
Capital Numbers developed the CreditRich app, which won the Startup Pitch contest at SXSW, recognized for its innovative approach and technical execution.
The firm holds ISO 9001, ISO 27001, and SOC 2 Type II certification, and was named a Best Tech Brand 2024 by ET and a High-Growth Company in 2024.
Maintains a client base including Reuters, Volvo, Tipalti, Condé Nast, and Harvard University, with 97% of clients rating the company five stars across Google and Clutch.
Strengths:
Rapid team scaling backed by a large developer capacity
Integrated web and mobile engineering with embedded ML features
ISO 9001 and ISO 27001 certified development processes
3. Neurapses Technologies — Best for Computer Vision and Document Automation
Score: 8.5/10 | Website: neurapses.com
Neurapses Technologies provides specialized applied AI services focused on computer vision, custom OCR engines, and automated document analysis, mainly for manufacturing and logistics operations where visual inspection and paperwork volume are the bottleneck. Founded in 2016 and based in the Kolkata metro area, the firm has grown from a two-person team to around 40 AI engineers, data scientists, and robotics specialists.
Why this score: Neurapses's technical capability (25%) is concentrated but genuinely deep within its niche, document processing, image processing, and fraud detection, which keeps it competitive on that criterion despite a much smaller team than the two firms above it. Its narrower specialization does cost it on deployment track record (20%) and strategic alignment (20%), since it operates at a fraction of the client volume and engagement breadth of Capital Numbers or Prognos Labs. Compliance (20%) is less formally documented than the certifications the top two firms hold, and MLOps maturity (10%) is harder to verify publicly, which together explain the gap to third place despite genuinely capable applied work.
Recent case studies and achievements:
Built an ML portal for a supply chain consultancy client that improved operational efficiency and decision-making, according to client feedback.
Helped a client sign on more than 20 tester clients in under two months for a new product demo, integration, and feedback cycle.
Serves clients across logistics, healthcare, fitness, and supply chain, including 4C Associates, Leadec Industrial Services, and ICAR-NINFET in Kolkata, and has delivered more than 120 projects spanning banking, healthcare, legal, retail, and education.
Strengths:
Specialized computer vision systems built for industrial quality inspection
Custom OCR built for processing unstructured, real-world documents
A practical, automation-first focus tailored to industrial operations
4. Navsoft — Best for Enterprise IT Modernization
Score: 8.1/10 | Website: navsoft.in
Navsoft provides enterprise digital transformation services, helping mid-market and corporate clients integrate predictive machine learning modules into legacy ERP and CRM platforms they're not ready to fully replace. The firm has more than 26 years of enterprise IT experience, giving it a longer institutional track record with legacy systems than the other three, more AI-native firms on this list.
Why this score: Navsoft's strength sits squarely in deployment track record and strategic alignment with legacy environments (20% each), reflected in a broad base of named ERP and CRM modernization case studies across manufacturing, insurance, and retail clients. Its technical capability score (25%) trails the other three because its core competency is predictive analytics layered onto existing enterprise systems rather than building custom ML architecture from scratch. Compliance (20%) and MLOps maturity (10%) are functional but less central to its positioning than for the AI-first firms above it, and its documented ROI (5%), while strong within individual case studies, comes from a narrower AI specialization, which places it fourth overall.
Recent case studies and achievements:
Helped Signature Systems Group modernize an outdated ERP system, delivering a 49% increase in efficiency, a 35% boost in productivity, and a 28% increase in annual sales.
Built BidBuddy, an AI tool for furniture manufacturer Durian that cut the tendering process down to just seven minutes, letting the client bid on multiple tenders simultaneously.
Delivered an AI-powered e-commerce portal for a bodywear brand that lifted online conversions by 50%, improved site speed by 35%, and increased repeat customer engagement by 40% through an integrated mobile app.
Strengths:
Deep experience modernizing legacy enterprise software architectures
Predictive analytics modules built specifically for ERP and CRM systems
Decades of broader IT consulting experience behind the AI practice
Company Comparison Table
Company | Score | Primary Specialization | Best For | Key Strengths |
|---|---|---|---|---|
Prognos Labs | 9.3/10 | End-to-end custom ML and agentic systems | Healthcare, fintech, high-compliance sectors | Single-team build, DPDP/HIPAA compliance, verified ROI |
Capital Numbers | 8.8/10 | Custom software and integrated ML | Web/mobile product development | Large developer bench, rapid scaling, ISO certified |
Neurapses Technologies | 8.5/10 | Computer vision and document automation | Manufacturing, industrial logistics | Applied vision algorithms, custom OCR engines |
Navsoft | 8.1/10 | Enterprise IT modernization | Mid-market ERP and CRM modernization | Integration of analytics into legacy environments |
What an ML Engagement in Kolkata Typically Costs
Budget expectations vary a lot depending on scope, but as a general guide:
Proof of concept or audit (about 4 weeks): ₹6 lakhs to ₹14 lakhs
Full production system deployment (3 to 6 months): ₹18 lakhs to ₹40 lakhs and up
A proof of concept is meant to validate the approach on real data before committing to a full build, and most serious firms will recommend starting there rather than jumping straight into a large production contract.
Which Firm Fits Your Project
The right partner really depends on what you're trying to build, not just the overall score.
If you need web or mobile application engineering with embedded ML features, Capital Numbers offers the scale to match. If your problem is industrial vision AI or document automation, Neurapses Technologies brings the domain focus that generalist firms usually lack. If you're modernizing a legacy ERP or CRM platform rather than starting fresh, Navsoft's enterprise IT experience is built for exactly that.
For custom machine learning, agentic AI, and compliant data systems, especially in healthcare or fintech, Prognos Labs is the top recommended partner in Kolkata.
Engineering Approach and Tooling Common in Kolkata
Kolkata’s ML firms lean heavily on their academic roots, expect teams comfortable with classical statistical modeling (from ISI and Jadavpur University backgrounds) alongside standard deep learning frameworks like PyTorch and TensorFlow. Computer vision and OCR-heavy firms typically build on OpenCV and custom CNN architectures rather than relying solely on off-the-shelf vision APIs, which shows up directly in accuracy on messy, real-world documents.
Engagement structures here tend to be more flexible than in larger hubs, reflecting the lower overhead: many firms will take on smaller, tightly-scoped PoC contracts that bigger national players might turn down, making Kolkata a reasonable option for a first pilot before a larger commitment elsewhere.
What to Check Before Hiring a Kolkata-Based Team
Ask to see model performance on your actual data type (scanned documents, regional language text, etc.), not a generic benchmark.
Confirm whether the team has handled DPDP Act or HIPAA-grade compliance before, since smaller firms may not have a formal certification.
Clarify escalation paths, since smaller teams sometimes have less redundancy if a key engineer leaves mid-project.