Pranav Jha

Machine Learning Engineer

Bengaluru, Karnataka, India8 yrs 5 mos experience

Most Likely To SwitchAI Enabled

Key Highlights

Architected a cloud-native AI framework at Ericsson.
Developed an unsupervised ML solution achieving 99% precision.
Created NLG microservices reducing manual reporting significantly.

Stackforce AI infers this person is a Machine Learning Engineer specializing in Telecommunications and Fintech industries.

Contact

pranavjha1993@gmail.com LinkedIn

Skills

Core Skills

Machine LearningGenerative AiMlopsData ScienceWeb Development

Other Skills

KubernetesSeldon CoreMLflowDaskArgo CDPrometheusGrafanaAlertmanagerTrivyGrypeKubesecKubeAuditSonarQubeCheckovScala

About

Hi, I'm Pranav Jha, a Machine Learning Engineer. I love solving real-world problems by building smart, scalable AI solutions from scratch. With nearly 8 years of experience, I’ve worked on data-driven projects that make businesses more efficient and accurate. At Ericsson, I designed a cloud-native AI framework used across the company, focusing on time-series forecasting, anomaly detection, and MLOps with tools like Kubernetes and PyTorch. At IBM, I built models for analytics, dashboards, and incident management, helping clients like State Street save time and costs. I’m skilled in Python, SQL, PyTorch, and frameworks like Scikit-learn and MLflow, with a knack for turning complex data into clear insights. In my free time, I enjoy coding personal projects to explore new AI techniques and improve my skills, often sharing my work on GitHub. I’m excited to keep using my expertise in math, coding, and machine learning to tackle challenging problems and deliver practical solutions. Connect with me to talk about AI, data, or new opportunities!

Experience

8 yrs 5 mos

Total Experience

4 yrs 2 mos

Average Tenure

4 yrs 4 mos

Current Experience

Ericsson

3 roles

Senior Machine Learning Engineer

Oct 2025 – Present · 7 mos · Bengaluru, Karnataka, India

Generative AIMachine Learning

Machine Learning Engineer III

Promoted

Jun 2024 – Oct 2025 · 1 yr 4 mos · Bengaluru, Karnataka, India

Cloud-Native AI Framework (MITO.ai)
Architected a cloud-native, no-code AI framework adopted as the foundation of Ericsson's AI strategy, enabling scalable, YAML-driven deployment of AI-native use cases as Kubernetes-compatible workloads across multiple vendors.
Designed an industrialized ML platform using Kubernetes, Seldon Core, MLflow, and Dask to support real-time analytics, cross-vendor interoperability, and seamless onboarding of use cases in IT, telecom, and network operations.
Built advanced ML workflows for time series forecasting, anomaly detection, trend analysis, RCA, and data quality checks, with production-grade drift detection to ensure model reliability.
Orchestrated end-to-end GitOps-based ML workloads using Argo CD with automated CI/CD pipelines to optimize deployment and iteration cycles.
Implemented comprehensive observability using kube-prometheus-stack, Prometheus, Grafana, and Alertmanager, delivering real-time insights into network anomalies across 20,000+ KPIs with dynamic dashboards.
Automated enterprise-grade security and compliance in CI pipelines using Trivy, Grype, Kubesec, KubeAudit, SonarQube, and Checkov for vulnerability detection and hardened deployments.
Engineered scalable data pipelines using Scala/Spark for high-volume time series processing, OpenSearch for storage, and MinIO for artifact management—delivering fault-tolerant, high-accuracy workflows.

KubernetesSeldon CoreMLflowDaskArgo CDPrometheus+14

Data Scientist II

Jan 2022 – Jun 2024 · 2 yrs 5 mos · Bengaluru, Karnataka, India

Intelligent Incident-Event Correlation
Built an unsupervised ML solution using K-Means clustering and NLP (Word2Vec, SpaCy) to automate incident and alarm mapping with existing incidents, reducing duplication across large volumes of real-time incident tickets and achieving 99% precision and 95% accuracy (SME-validated).
Built Seldon Core microservices and Dask-based visualization dashboards to deploy and monitor the ML solution, replacing manual analysis processes with automated insights.

K-Means clusteringNLPWord2VecSpaCySeldon CoreDask+2

Ibm

3 roles

Data Scientist

Promoted

Jan 2019 – Jan 2022 · 3 yrs

Augmented Analytics & NLG
Built tornado template-based NLG microservices automating prescriptive dashboard insights and reducing manual reporting across client accounts.
Created a Python library for auto-generating tornado templates from English text, enabling portable domain-specific narratives with accelerated development cycles.
Monitoring & Anomaly Detection
Developed ensemble anomaly detection models (Isolation Forest, Extreme Value Theorem) with Pareto-based noise filtering, improving incident detection accuracy across multiple clients.
Built Python-based Kibana Rollup jobs for data aggregation, reducing storage costs by 20% and eliminating licensing fees.
Time Series & Network Analytics
Implemented AutoARIMA forecasting with pmdarima for network/storage resource optimization and cost reduction.
Built Python exporters for Juniper/Aruba networks using Mist/CSO APIs with real-time Prometheus/Grafana monitoring dashboards.
Processed multi-source data (Elasticsearch, Kafka, Cassandra) using Python/PySpark for Kibana visualization insights.
Incident Management & SLA Prediction
Designed ML and rule-based proactive incident management engine with defect classification, reducing ticket counts by 65% for clients like State Street.
Developed random forest SLA prediction models enhancing contract accuracy through defect and support metrics correlation.
Explainable AI & Dashboard Solutions
Implemented explainable AI solutions maintaining high prediction accuracy while improving model transparency.
Built Dash/Plotly web dashboard for IBM cloud billing insights, serving 200+ users.

NLGPythonKibanaElasticsearchKafkaCassandra+5

Python and Web Application Developer

Promoted

Nov 2017 – Dec 2018 · 1 yr 1 mo

Revamped a web app for an Italian bank using Python, HTML, and JavaScript, increasing user engagement by 30%.
Automated Excel-based business logic with Python scripts, saving 20 hours weekly.

PythonHTMLJavaScriptWeb Development