Syed Ashfaq
Machine Learning Engineer
7+ years exp
C++
SQL
Node JS
TypeScript
JavaScript
C
Python
Model Architecture & Training: 4/10
Active 3 days ago
+91 (8438) 524104 Invite to interview
Message
Download CVCV
Overview
Technical skills
Timeline
Roles
Overview
Computer vision ML engineer (mid-level) who reproduces research models and implements training pipelines in PyTorch. The strongest proven skill is implementing and integrating SE modules into ResNet and Inception backbones, demonstrated by the se_module.py, se_resnet.py and se_inception.py implementations. The public code lacks structured experiment tracking, comprehensive data augmentation pipelines, tests, and deployment/serving infrastructure.
Technical skills
C++
SQL
Node JS
TypeScript
JavaScript
C
Python• Middle • 7y+
C++
PyTorch C++
TensorFlow C++
Python
Dask
FastAPI
Databases
Databricks
Milvus
MySQL
Neo4j
Pinecone
PostgreSQL
Redis
AI/ML
CUDA
CUDA Toolkit
Data Augmentation
Fine-tuning
Flash Attention
GAN
JAX
Keras
LangChain
LLM
LoRA
MLFlow
NLP
NLTK
NumPy
Pandas
PyTorch
QLoRA
RAG
Scikit-learn
TensorFlow
Transformers
Spark
PEFT
Deep Learning
Computer Vision
Torchvision
Datasets
AI/ML
LangSmith• 3y+
DevOps
AWS
CI/CD
Docker
GCP
Kubernetes
Vector
Rest API
Azure• 7y+
Analytics
Matplotlib
Power BI
Seaborn
Tableau
Timeline
Machine Learning Software Engineer
•
Middle
I-Exceed Technologies Pvt Ltd
•
Full-Time
Built NLP solutions for business use cases, including intent and named-entity recognition, text classification, and sentiment analysis. Developed multi-stage NLP pipelines using Azure services and created retrieval and question-answering components using document parsing and RAG patterns. Implemented onboarding-related tools including OCR-based extraction and face detection/matching workflows, and added data augmentation plus ML pipeline design with Airflow.
Python
Azure
LangSmith
Vellore Institute of Technology (VIT)
Bachelor's Degree •
Electrical and Electronics Engineering
Middle AI/ML Engineer
Confidence: Medium ML Engineer
Computer vision ML engineer (mid-level) who reproduces research models and implements training pipelines in PyTorch. The strongest proven skill is implementing and integrating SE modules into ResNet and Inception backbones, demonstrated by the se_module.py, se_resnet.py and se_inception.py implementations. The public code lacks structured experiment tracking, comprehensive data augmentation pipelines, tests, and deployment/serving infrastructure.
Model Architecture & Training
4/10
How well models are designed and trained
Clear, working PyTorch implementations of SE blocks and SE-ResNet variants plus training scripts; demonstrates correct model-building patterns and integration of SE modules into ResNet/Inception.
Evidence
Computer-Vision-Projects/SENet1/senet/se_module.py: SELayer class
Computer-Vision-Projects/SENet1/senet/se_resnet.py: SEBasicBlock, SEBottleneck, se_resnet50 function
Computer-Vision-Projects/SENet1/senet/se_inception.py: SEInception3 integration with torchvision.models.Inception3
Data Pipeline & Feature Engineering
2/10
How data is prepared for models
Minimal data-pipeline work is present; dataset loading is delegated to homura DATASET_REGISTRY with no visible augmentation, preprocessing, or leakage handling logic.
Evidence
Computer-Vision-Projects/SENet1/cifar.py: DATASET_REGISTRY('cifar10') usage
Computer-Vision-Projects/SENet1/imagenet.py: DATASET_REGISTRY('imagenet') usage
Experimentation & Evaluation
2/10
How results are measured and tested
Basic experiment/eval primitives are used (accuracy callbacks, loss reporting, TensorBoard), but there is no structured experiment tracking, baselines/ablation harness, or reproducible experiment metadata.
Evidence
Computer-Vision-Projects/SENet1/imagenet.py: callbacks list includes AccuracyCallback, LossCallback, reporters.TensorboardReporter
Computer-Vision-Projects/SENet1/cifar.py: reporters.TQDMReporter and callbacks.AccuracyCallback usage
MLOps & Deployment
2/10
How models are shipped to production
Some awareness of distributed training and checkpoints (init_distributed, WeightSave), but no serving, model versioning, CI/CD, or drift/monitoring infrastructure is present.
Evidence
Computer-Vision-Projects/SENet1/imagenet.py: init_distributed() and is_distributed() usage
Computer-Vision-Projects/SENet1/imagenet.py: callbacks.WeightSave('.') for checkpointing
Computational Efficiency
3/10
How efficiently computing resources are used
Shows some computational considerations such as LR scaling with batch size and using adaptive pooling; no advanced GPU/quantization/batching optimizations are implemented.
Evidence
Computer-Vision-Projects/SENet1/imagenet.py: optimizer lr set as 0.6 / 1024 * args.batch_size (LR scaling)
Computer-Vision-Projects/SENet1/senet/se_resnet.py: use of nn.AdaptiveAvgPool2d in model definitions
Research Depth & Innovation
3/10
Depth of research and new ideas
Paper reproduction quality is solid: SE module and its integration into multiple backbones indicates research-implementation skill, but there is no evidence of novel algorithms, ablation studies, or published results.
Evidence
Computer-Vision-Projects/SENet1/senet/se_module.py: SELayer implementation
Computer-Vision-Projects/SENet1/senet/se_inception.py: adding SELayer modules into Inception blocks
Expertise
Computer Vision & Image Analysis• Middle
Technologies
Deep Learning
Python• Middle • 7y+
SQL
C++
MySQL
PostgreSQL
Redis
Rest API
LangChain
Neo4j
Spark
GCP
Databricks
Milvus
Pinecone
Flash Attention
LoRA
CUDA Toolkit
MLFlow
FastAPI
Fine-tuning
Scikit-learn
JAX
Computer Vision
NLP
LangSmith• 3y+
PEFT
QLoRA
Azure• 7y+
Datasets
CI/CD
Transformers
TensorFlow
Pandas
NumPy
Keras
PyTorch
AWS
Docker
Kubernetes
LLM
RAG
NLTK
TensorFlow C++
PyTorch C++
Dask
Torchvision
Vector
GAN
Data Augmentation
CUDA
Recommendations
- Develop end-to-end training pipelines with explicit data augmentation, preprocessing and validation splits to harden experiments.
- Add experiment tracking (W&B or MLflow), reproducible configs, and automated eval/ablation scripts to make results comparable and repeatable.
- Implement unit/integration tests for model components and add CI to protect refactors and verify training scripts.
- Work on inference/serving artifacts (serialization, example inference scripts, simple REST endpoint) and basic performance benchmarks (latency/throughput).
Repositories
The developer's experience in this domain has been verified based on AI analysis of the following repositories:
Intern Data Scientist
Confidence: Low Generalist
An early-career generalist with an interest in AI/ML and library exploration, currently at an intern level. The strongest proven skill is familiarity with AI/ML topics as shown by README entries describing AI and machine learning tools such as Obsei and Model Store. There is no original production code, notebooks, data pipelines, tests, or CI visible to support a higher engineering or data-science rating.
Statistical Rigor
Correct use of statistics
Not evidenced in public code
Data Wrangling & Cleaning
Preparing and cleaning data
Not evidenced in public code
Exploratory Analysis & Visualization
Exploring and visualizing data
Not evidenced in public code
Predictive Modeling
Building models that predict
Not evidenced in public code
Business Insight & Impact
Turning analysis into business value
Not evidenced in public code
Reproducibility & Notebook Hygiene
Clean, repeatable analysis
Not evidenced in public code
Technologies
AI/ML
Tableau
Power BI
Seaborn
Matplotlib
Recommendations
- Build a small, self-contained ML notebook project with data collection, cleaning, EDA, clear hypothesis notes, and a documented train/test split to demonstrate end-to-end workflow.
- Develop at least one reproducible script or pipeline (with tests and pinned environment) that ingests data, performs transformations, and outputs a validated dataset to show engineering ownership.
- Publish a short project including unit tests, README with architecture decisions, and a tracker (DVC or dataset snapshot) so skills can be assessed from concrete, human-authored artifacts.
Repositories
The developer's experience in this domain has been verified based on AI analysis of the following repositories:
Intern Backend Developer
Confidence: Low Generalist
Early-career backend developer at an intern level who appears to be exploring and documenting libraries and tooling. The only proven artifact is non-code documentation in README files, so no implemented backend feature demonstrates a concrete technical strength. There is no public server-side code, tests, migration history, CI configuration or operational instrumentation to evaluate system design or production-readiness.
API Design
How well APIs are designed
Not evidenced in public code
Data Layer & Database
Working with databases
Not evidenced in public code
Scalability & Performance
Handling load and speed
Not evidenced in public code
System Architecture
Overall system structure
Not evidenced in public code
Security & Auth
Protecting data and access
Not evidenced in public code
Reliability & Observability
Stability and monitoring
Not evidenced in public code
Recommendations
- Build a small, documented backend service (REST or GraphQL) with at least one database-backed resource, including migrations and transactional boundaries.
- Add automated tests and a CI pipeline that runs them, and include structured logging and basic metrics to demonstrate observability.
- Implement authentication and input validation for the service and add a README describing API contracts, deployment steps and scaling considerations.
Repositories
The developer's experience in this domain has been verified based on AI analysis of the following repositories: