Every planning decision, backed by spatial evidence.

The programme

AGT Intelligence leverages AI and geospatial data to accelerate high-stakes decision-making for governments, enterprises, and individuals. By combining advanced analytics, remote sensing, and domain expertise, we deliver data-driven intelligence that influences real-world outcomes. This role sits at the core of our technical program, focusing on the extraction and interpretation of high-resolution aerial, drone, and satellite imagery, alongside LiDAR and elevation data, to create reliable, defensible municipal records.

What you will do

• You will own the model architecture and the overarching accuracy standards for every feature the platform extracts. Because these outputs serve as official records that governments act upon, you will define strict accuracy contracts, dictating how model performance is measured, verified on unseen areas, and signed off before deployment.
• Day-to-day operations include exploring new research methodologies, prototyping and productionising models, reviewing training data pipelines, mentoring technical teams, and collaborating with engineering to ensure your architectures scale reliably in real-world environments.

What you will need

• Strong foundation in Computer Science, including data structures, algorithms and system design.
• Proficiency in programming languages commonly used in AI and computer vision (Python, C++ or similar).
• Hands-on software development experience, including version control, testing and deploying production-grade code.
• Exposure to distributed systems, cloud platforms or data pipelines.
• Demonstrated expertise in computer vision, remote sensing and geospatial analytics, including work with aerial or satellite imagery.
• Experience with deep learning frameworks (PyTorch, TensorFlow) and familiarity with MLOps practices.
• Experience defining how models are evaluated, including testing on genuinely unseen areas, and standing behind the results.
• Advanced degree (Master's or PhD) in Computer Science, Electrical Engineering, Remote Sensing or a related technical discipline.
• Ability to lead research projects, mentor technical teams and communicate complex concepts clearly to non-technical stakeholders.

Tools & skills

• PyTorch
• Computer Vision
• Remote Sensing
• Deep Learning
• Image Segmentation
• Photogrammetry
• LIDAR
• Machine Learning
• Python
• Rasterio
• MLflow
• Image Interpretation
• Polygon Annotation
• Dask

The programme

Operating within AGT Intelligence's broader mission to provide rapid, data-driven insights to public and private sectors, this engineering stream is dedicated to turning high-resolution aerial imagery and elevation data into structured, highly accurate map layers. The focus is on solving complex geometric challenges in dense urban environments across India, ensuring that inference works flawlessly at a city-wide scale.

What you will do

• You will build, train, and ship robust deep learning models, taking them from experimental phases directly into production. Your responsibilities include diagnosing model failures, refining training data and labels, and converting raw model outputs into clean, validated vector geometry usable by downstream systems.
• You will optimize hardware performance, process large images in tiles, and work closely with platform teams to guarantee that your pipelines are fully reproducible and that results can be verified by peers.

What you will need

• Strong foundation in Computer Science, including data structures, algorithms and software design.
• Proficiency in Python and hands-on experience with deep learning frameworks, primarily PyTorch.
• 4-7 years building and shipping deep learning vision models, with at least one year on aerial, satellite or other overhead imagery.
• Practical experience with segmentation and detection architectures, and with multi-sensor or multi-channel model inputs.
• Familiarity with geospatial and imaging libraries (OpenCV, NumPy, rasterio, geopandas, GDAL) and with processing large images in tiles.
• Experience taking geometry from model output through to clean, validated vector data.
• Working knowledge of Docker, Git, experiment tracking and GPU optimisation.
• Bachelor's or Master's degree in Computer Vision, Computer Science, Remote Sensing, Geoinformatics or a related field.
• Methodical debugging, collaborative working style and a commitment to reproducible engineering.

Tools & skills

• Computer Vision
• PyTorch
• LIDAR
• MLflow
• Deep Learning
• SAM Pre-labeling
• Image Interpretation
• Remote Sensing
• Python

The programme

AGT Intelligence supports high-stakes decision environments by managing complex geospatial insights. This backend initiative focuses on constructing the foundational data platform that ingests, versions, and serves massive volumes of geospatial data. This system acts as a governed system of record for survey data, drone measurements, and AI-extracted features, requiring a pristine historical record where ground captures are versioned rather than overwritten.

What you will do

• You will architect and own the core database schemas, APIs, and map services, ensuring the platform remains fast and secure as it scales across multiple cities. You will design time-aware data models that maintain strict provenance and historical accuracy for every measurement.
• Daily tasks include tuning spatial queries, building robust ingestion pipelines, implementing role-based access control and audit logs, and managing containerized on-premises deployments for government clients.
• You will also set the technical standards for the wider backend engineering team.

What you will need

• Strong foundation in Computer Science, including system design, data modelling and algorithms.
• Expert Python, with production experience of a modern web framework such as FastAPI, Django or Flask.
• Advanced PostgreSQL, ideally with PostGIS: schema design, indexing, query optimisation and handling of large datasets.
• 7+ years in backend engineering, including 3+ years architecting data platforms, spatial databases or regulated enterprise systems.
• Experience designing systems that keep history and provenance: versioned or time-aware data models.
• Familiarity with geospatial standards and formats, map and tile services, and object storage.
• Working knowledge of Docker, CI/CD, database migrations, access control and audit logging.
• Bachelor's or Master's degree in Computer Science, Geoinformatics, Software Engineering or a related field.
• Ability to defend architectural decisions clearly, document them well and mentor engineers.

Tools & skills

• PostgreSQL
• PostGIS
• FastAPI
• Data Architecture
• REST APIs
• Docker
• SQL
• System Design

The programme

As part of AGT Intelligence's mission to accelerate decision-making, this program builds the intelligence layer atop our geospatial platform. This is not an open-ended chatbot; it is a highly constrained, agentic workflow and retrieval system designed to synthesize spatial, historical, and statutory data into clear, traceable reports and drafted documents for official government use.

What you will do

• You will design and deploy tool-calling pipelines and retrieval systems over structured and unstructured data sources, ensuring every generated answer cites the exact municipal record, survey, or rule it relies on.
• Your primary deliverable before shipping any agent is building a rigorous evaluation harness to gate releases.
• You will also implement safeguards against data leakage, automate data extraction from scanned statutory documents, manage latency, and deploy self-hosted models to comply with strict Indian data residency requirements.

What you will need

• Strong foundation in Computer Science, including data structures, algorithms and API design.
• Proficiency in Python, with production experience of FastAPI or a similar framework, SQL and asynchronous programming.
• 4-8 years in software engineering, including 2+ years building and deploying LLM, RAG or agentic systems used by real users in production.
• Practical experience with agent frameworks, tool calling, structured outputs and multi-step workflows.
• Experience with retrieval-augmented generation: vector search, hybrid retrieval, embeddings and reranking.
• Experience building evaluation frameworks for AI systems, and using them to decide what ships.
• Awareness of AI security and privacy concerns, including prompt injection, data leakage and handling of personal data.
• Exposure to self-hosted or local model inference is an advantage.
• Bachelor's or Master's degree in Computer Science, Data Science, AI or a related discipline.

Tools & skills

• Large Language Models (LLM)
• Retrieval-Augmented Generation (RAG)
• Python
• AI Agents
• Prompt Engineering
• FastAPI
• Vector Databases
• PostgreSQL
• Machine Learning
• Docker

The programme

AGT Intelligence leverages AI and geospatial data to accelerate high-stakes decision-making for governments, enterprises, and individuals. This role bridges the gap between raw spatial data engineering and applied analytical science, focusing on extracting actionable intelligence from municipal, environmental, and temporal datasets. The work directly supports automated land intelligence, urban sprawl analysis, and climate risk assessments, transforming complex raster and vector data into defensible, governed records.

What you will do

• You will lead the core raster and vector data science initiatives, applying advanced statistical modeling and machine learning to massive spatial datasets. Day-to-day, this involves performing deep spatiotemporal analysis on dense raster stacks and complex vector geometries to model real-world phenomena such as urban sprawl, infrastructure deformation, and climate risk exposure.
• You will design algorithms to process spatial tensors, build complex mathematical processing pipelines for multi-sensor data (including radar remote sensing and persistent scatterer interferometry), and engineer spatial features for downstream machine learning models.
• The role demands solving complex spatial data science problems by manipulating high-dimensional data using modern cloud-native formats like Zarr, GeoParquet, and Apache Parquet to deliver statistically rigorous, mathematically proven spatial intelligence.

What you will need

• An advanced degree (Master's or PhD) in Data Science, Geological Sciences, Geoinformatics, or a highly related quantitative discipline.
• Rigorous, hands-on experience in applied spatial data science, scripting heavy vector and raster analytics using the Python ecosystem (GeoPandas, Rasterio, NumPy).
• A deep understanding of raster math, spatial autocorrelation, machine learning on spatial tensors, and coordinate reference systems.
• Experience with advanced geospatial modeling techniques, such as Synthetic Aperture Radar (SAR) processing or automated spatiotemporal change detection, is highly preferred.

Tools & skills

• PyTorch
• Computer Vision
• Dask
• QGIS
• SAR
• LIDAR
• GDAL
• Rasterio
• MLflow
• NetCDF
• Deep Learning
• Spatiotemporal Modeling

The programme

AGT Maps is building a next-generation Municipal Geospatial Intelligence Platform. This platform fuses ultra-high-resolution aerial capture (nadir, oblique, and LIDAR) with multi-modal satellite imagery to automate the extraction of urban features and drive immediate civic action. Operating on an aggressive delivery window, the program focuses on deploying scalable computer vision architectures to serve core modules like Property Intelligence, Illegal Activity Detection, Water & Utility Intelligence, and bi-temporal Change Detection.

As the Data Engineer (raster/vector pipelines) at AGT Maps, you are the backbone of the platform's spatial data architecture. You will own the critical data flows from raw sensor ingest to clean, queryable municipal databases: ingest, tiling, vectorisation, and topology, ensuring the multi-task AI models have perfectly aligned raster grids to infer from, and that the resulting outputs are transformed into flawless, topologically sound vector datasets mapped directly to the AMRUT schema.

What you will do

• Ingest raw image frames, GNSS/IMU logs, and LIDAR from aerial capture teams, running automated checks for completeness, overlap, and exposure before accepting blocks into the pipeline.
• Build and manage generation pipelines for Cloud-optimised GeoTIFFs, establishing fixed tile grids with strict overlap aligned across historical and current epochs for accurate change detection.
• Engineer high-volume distributed spatial data streams utilizing Dask processing to handle chunked, multi-dimensional raster arrays efficiently without memory bottlenecks.
• Develop automated raster-to-vector conversion pipelines, executing polygon simplification, rigorous topology cleanup, and geometric snapping.
• Map all output vector layers cleanly to the AMRUT GIS schema and populate required attributes within a centralized PostGIS database.
• Manage LIDAR and photogrammetric point cloud processing pipelines utilizing PDAL, Entwine, and LAZ formats.
• Implement and maintain a STAC (SpatioTemporal Asset Catalog) to track capture epochs, manage data provenance, and serve the change-detection engine.
• Collaborate closely with the AI and CV teams to ensure seamless data handoffs between the photogrammetry outputs, the inference engine, and the final GIS database structure.

What you will need

• Bachelor's or Master's degree in Computer Science, Geoinformatics, Geospatial Engineering, or a related highly quantitative field.
• 3-5 years of dedicated, full-time experience in geospatial data engineering, focusing heavily on automated spatial pipelines and large-scale data processing.
• Expertise in spatial databases, specifically PostgreSQL with PostGIS, for heavy vector manipulation, indexing, and executing topology rules.
• Advanced programming skills in Python, with proven experience engineering distributed spatial data streams using Dask processing to overcome memory limits on massive arrays.
• Strong proficiency with core geospatial libraries (GDAL, Rasterio, Shapely, GeoPandas) and point cloud utilities (PDAL).
• Deep experience generating and managing Cloud-Optimised GeoTIFFs (COG) and working with STAC metadata specifications.
• Familiarity with spatial visualization and desktop analysis tools like QGIS to validate data pipelines and schema mappings.
• Highly motivated, proactive work style with a strong attention to detail regarding data quality, spatial topology, and geometric accuracy.

Tools & skills

• QGIS
• LIDAR
• GDAL
• Rasterio
• NetCDF
• PyTorch

The programme

Our platform is building a next-generation Municipal Geospatial Intelligence Platform, fusing ultra-high-resolution aerial capture (nadir, oblique, and LiDAR) with multi-modal satellite imagery to automate the extraction of urban features and drive immediate civic action. The programme focuses on deploying scalable computer vision architectures to serve core modules like Property Intelligence, Illegal Activity Detection, Water & Utility Intelligence, and bi-temporal Change Detection.

As a GIS Feature Extraction Specialist/Annotator, you will be at the frontline of creating foundational training datasets for our AI models. Utilizing AI-assisted tools like the Segment Anything Model (SAM), you will perform precise polygon and bounding box labeling on high-resolution satellite and aerial imagery, converting complex urban landscape features into high-quality vector inputs.

What you will do

• Perform high-precision vector labeling, polyline tracing, and bounding box annotations on satellite and aerial imagery using platforms like CVAT and Label Studio.
• Leverage SAM-assisted (Segment Anything Model) pre-labeling tools to accelerate feature extraction across building footprints, roads, water bodies, and urban vegetation.
• Identify, classify, and delineate complex urban infrastructure elements across varying image resolutions and viewing angles.
• Perform initial self-checks on vector boundaries, topology, and attribute tags to meet strict accuracy thresholds prior to QA/QC handoff.
• Collaborate closely with QC Reviewers and Annotation Managers to adapt to updated tagging guidelines, edge-case protocols, and throughput targets.

What you will need

• Education: Diploma or Bachelor's degree in Geoinformatics, Geography, Computer Applications, or a related field.
• Computer & GIS literacy: good computer proficiency with a basic understanding of maps, satellite imagery, or spatial data representation.
• Labeling experience: hands-on familiarity with data annotation platforms (CVAT, Label Studio, Kili, or Labelbox).
• Attention to detail: excellent visual interpretation skills with the ability to distinguish subtle geographic and structural features in aerial imagery.
• Tool familiarity: basic exposure to GIS software (e.g. QGIS) or AI-assisted interactive segmentation tools is a plus.

Tools & skills

• SAM Pre-labeling
• Image Interpretation
• Polygon Annotation
• Computer Vision
• QGIS
• LIDAR
• GDAL
• Rasterio
• Vector Digitizing
• Raster Audit

The programme

As a Geospatial Data Science Intern, you will help bridge the gap between complex earth observation data and actionable intelligence. You will work directly with remote sensing datasets to assist in developing spatiotemporal models and automated spatial analysis workflows.

What you will do

• Assist in processing and analyzing large-scale remote sensing datasets (e.g. Sentinel-1) for land deformation, flood inundation, and infrastructure risk profiling.
• Run analytical workflows and spatial queries using Google Earth Engine and QGIS.
• Help build and refine automated modeling pipelines for property-level and asset-level climate risk projections.
• Collaborate on technical documentation, mapping outputs, and cartographic visualizations.

What you will need

• Strong foundational knowledge of geographic information systems and spatial data structures (raster/vector).
• Proficiency in Python and spatial libraries (e.g. GeoPandas, Rasterio, Shapely).
• Familiarity with Google Earth Engine, QGIS, or similar geospatial frameworks.
• A background in Data Science, water resources, remote sensing, or a related field is a major plus.

Tools & skills

• QGIS
• Rasterio
• PyTorch
• NetCDF

The programme

As a Geospatial Data Engineer Intern, you will focus on the architecture and infrastructure that keeps our data flowing seamlessly. You will help build out the backbone of our data operations, ensuring that high-volume streams are processed efficiently and accurately.

What you will do

• Assist in designing and maintaining robust data pipelines to extract, transform, and load (ETL) multi-variate datasets.
• Implement and optimize large-scale distributed workloads using Dask processing for faster model execution.
• Integrate third-party APIs and automate workflow syncs across various cloud environments and databases.
• Write clean, maintainable code to support local hardware acceleration and pipeline optimization.

What you will need

• Solid programming skills in Python and an understanding of distributed computing concepts.
• Familiarity with data stream engineering, API integrations, and database management.
• Experience or strong interest in parallel processing frameworks (like Dask) and cloud-native workflows.
• A problem-solving mindset with the ability to troubleshoot bottlenecks in data ingestion and storage.

Tools & skills

• PyTorch
• Dask
• QGIS
• SAR
• NetCDF
• Deep Learning