Senior Data Scientist

MonoEdge; Pune District; On-site; INR 5,500,000 - 7,500,000; Full time

Job summary

MonoEdge is seeking a data scientist to design and operate the analytics backbone for an industrial optimization platform in India. You will connect plant data streams with downstream insights, quantify ROI, and collaborate across domains including production, sales, and finance.

You will own data architecture, label data for modeling, and validate models with ground-truth experiments. Sentences emphasize practical impact, cross‑lingual plant communication in Hindi/Marathi, and hands-on work in

Qualifications

• 4+ years of data science experience with industrial or operational data in manufacturing, energy, or logistics.
• Strong applied statistics: regression, hypothesis testing, experimental design, confound handling with observational data.
• Proficient in Python with data science stack and SQL; able to structure a real codebase, not just notebooks.

Responsibilities

• Own the data architecture for the core wedge product: schema design, ingestion, joining heterogeneous sources, labeling protocol.
• Run domain workshops with plant QC and operations managers to translate tacit knowledge into a modelable schema.
• Build and validate the core decision model that sits on top of perception features from the CV pipeline.
• Design ground-truth collection and validation framework for advisory and closed-loop deployment.

Skills

Industrial data analytics

Time-series modeling

Causal inference

Communication in Hindi/Marathi

Ground-truth design

Education

Bachelor's degree in a quantitative field

Tools

Python

SQL

Pandas

scikit-learn

Statsmodels

Job description

MonoEdge develops an advanced intelligence layer for the Indian mid-market manufacturing sector. We provide data-driven insights and strategic recommendations to plant leadership and operations supervisors across diverse linguistic contexts, including English, Hindi, and Marathi.

As a technically-driven, bootstrapped organization, we are defining a new category in industrial optimization.

What you'll work on

• Yield, quality, and OEE modeling — connecting upstream process data to downstream outcomes (rejects, downgrades, throughput, energy).
• Industrial time-series analysis — PLC streams at 10–100 Hz, lab data at hour or shift frequency, ERP data at day frequency. Making them queryable and analytically joinable.
• Causal inference on observational data — most decisions in a plant happen without an A/B framework. Designing analyses that survive confounding, drift, and operator override.
• Ground-truth design and experimental design — generating labels where none exist, getting expert judgment efficiently, and validating models in production.
• ROI quantification — every model output must be defensible in rupees. You will own that defensibility.
• Cross-domain analytics — the platform connects production data with sales, procurement, inventory, and finance. You will help shape how these are joined and what insights they unlock together.

What you'll do

• Own the data architecture for the core wedge product: schema design, ingestion, joining heterogeneous sources, labelling protocol.
• Run domain workshops with plant QC and operations managers to translate their tacit knowledge into a labelable, modelable schema.
• Build and validate the core decision model that sits on top of perception features from the CV pipeline.
• Design the ground-truth collection — including operator judgment elicitation — and the validation framework for advisory and, later, closed-loop deployment.

What you bring

• 4+ years of data science experience with at least 2 years working with industrial or operational data — manufacturing, energy, logistics, supply chain. Not consumer or web product analytics.
• Strong applied statistics: regression, hypothesis testing, experimental design, confound handling, working with observational data. You can defend a model's claim in front of a sceptical operator.
• Strong Python: pandas, scikit-learn, statsmodels, plotting libraries. Comfortable in SQL. Can structure a real codebase, not just notebooks.
• Cost-benefit thinking: you know when a 92% model that ships beats a 96% model that doesn't. You think about decision quality, not metric quality.
• Comfort with messy data: missing values, inconsistent units, undocumented sources, partial coverage. You fix it patiently rather than complain about it.
• Plant-floor empathy: you are willing to spend a week in a manufacturing plant, talking to a shift supervisor in Hindi or Marathi, to understand what the data actually means before you model it.

Nice to have

• Causal inference toolkit: DAGs, propensity scoring, instrumental variables, difference-in-differences.
• Time-series at industrial frequencies: PLC data, sensor streams, alignment, resampling, drift handling.
• Familiarity with manufacturing concepts: yield, OEE, first-time-right, kWh/MT, throughput, ageing inventory, working capital.
• Experience working alongside computer vision pipelines, taking perception outputs into downstream models.

How we work

Reporting to the Founder, you will steward the analytical and statistical framework of the platform. You are expected to design robust solutions, provide critical technical oversight, and maintain ownership of both core and exploratory analytics to drive the organization’s strategic spine.

We operate with agility and a rigorous focus on EBITDA-level impact. Our performance is validated by real-world financial improvements rather than isolated metric performance.