Hello There
An Associate Machine Learning Engineer role has opened up at a Salt Lake Sector V employer in Kolkata, full-time and based on-site. If you've spent a couple of years turning messy data into models that actually ship, this is worth a closer look. The team here works across a handful of internal products rather than one narrow use case, so the modeling problems change more often than they would at a company with a single flagship product. One month might be about predicting customer churn, and the next about forecasting inventory needs for a completely different part of the business. That variety is part of the appeal for someone still figuring out which corner of machine learning they want to specialize in long-term.
The Day-to-Day
The work moves between building things and watching how they behave once they're live.
- Design and fit predictive models around specific business problems
- Dig through large datasets to pull out patterns worth acting on
- Get finished models running in the live systems that depend on them
- Watch how those models hold up once real traffic starts hitting them
A model that scores well in a notebook doesn't always hold up once it meets live data, and figuring out why a live model's accuracy has quietly slipped over three weeks is often more interesting than building the original model. Part of the job is also translating what a stakeholder actually needs into something measurable and trainable, since the request that comes in is rarely already framed as a clean modeling problem.
What You'll Bring
The academic baseline is a four-year degree in computer science, statistics, or a similarly numbers-heavy field. What matters more day-to-day is two years of real experience actually shipping machine learning work, especially if that experience includes taking a model past the prototype stage and into something people actually use.
- Python
- TensorFlow or PyTorch
- Machine learning algorithms
- SQL
Nice to have but not required: some exposure to a cloud platform's ML tooling, or a project where you had to explain a model's output to someone who wasn't technical. Both come up often enough in this role that having them already will save some ramp-up time. Familiarity with version control for models and datasets, not just code, also helps, since reproducing an old result six months later is a common enough request that it's worth being ready for.
Pay and Perks
This position pays up to ₹91,500 a month. Salt Lake Sector V has become one of the busier pockets for machine learning engineer jobs in Kolkata over the last couple of years, and the on-site expectation here comes with a few things meant to offset the commute.
- Health insurance
- Paid time off
- Provident fund contributions
- On-site cafeteria access and cab or commute support
Naukri Mitra often hears that commute support matters more to candidates than it looks on paper, especially for anyone coming from the other side of the city during Kolkata traffic.
The Team
This is a small associate cohort working alongside a couple of senior engineers who review model design decisions before anything reaches production. Expect regular back-and-forth on why a particular feature was included or dropped, not because anyone's being precious about the modeling choices, but because a model that's hard to explain later is a model that's hard to debug later too. Standups run short and focus on blockers rather than status updates for their own sake, and most of the deeper design conversations happen outside of them entirely. New associates typically pair with a senior engineer on their first couple of projects rather than being handed something to figure out alone, which shortens the learning curve considerably.
Where the Work Gets Harder
Not every week is clean experimentation. Some sprints go entirely toward cleaning a dataset that has more gaps than expected, or retraining a model after a business team quietly changes what a key input actually means. That kind of work rarely shows up in a portfolio, but it's most of what keeps a production model trustworthy over time. An associate who's willing to sit with that unglamorous part of the job tends to become the person others rely on within a year, well before anyone officially calls them senior. There will also be stretches when a model that worked fine for months suddenly starts drifting for reasons that take a few days of digging to pin down, and that kind of troubleshooting is treated as a normal part of the role rather than an emergency exception.
Growth From Here
Two years in, engineers from this kind of associate seat typically move toward a mid-level ML role, specialize in a particular modeling domain like recommendation systems or forecasting, or shift toward ML infrastructure if the deployment side turns out to be the more interesting half of the job. Which direction fits best usually becomes clear during the first year, once you've seen which parts of the work you keep gravitating toward on your own. Someone who enjoys the debugging and monitoring side more than the initial model-building often ends up drawn toward infrastructure roles later, and that's a perfectly reasonable path even if it wasn't the plan going in.
Applying
Send your resume along with a link to any model you've built end-to-end, from notebook to production; even a personal project counts if it went beyond a proof of concept. Interviews here spend real time on a take-home or live coding exercise involving an actual dataset, so it helps to be comfortable talking through your reasoning out loud rather than just presenting a finished result. Candidates who ask specific questions about the datasets they'd be working with tend to stand out more than those who ask only about the tech stack. If you're currently commuting from elsewhere in Kolkata, it's worth asking about the cab support upfront, since eligibility can depend on which part of the city you're coming from.
