[ LIVE_FEED · ONLINE ] BENGALURU, IN
// VISHWAS N — PORTFOLIO · V2026

VISUAL
INTELLIGENCE
& EDGE THREAT
DEFENSE_

› Vishwas N — final-year AI/ML engineer shipping computer-vision systems from data ingest to edge inference. Currently building edge-deployed classifiers, behavioral malware detectors, and large-scale annotation pipelines.

VIEW OPERATIONS ↘ OPEN CHANNEL
94.1%
TOP-1 ACCURACY
55K+
IMAGES TRAINED
1,046
SPECIES CLASSES
<1s
EDGE INFERENCE
AIML ENGINEER +ML ENGINEER +CYBERSECURITY ENTHUSIAST +
01 — ABOUT

RUN.
BIO.SH

vishwas@portfolio:~$
› cat bio.txt
Final-year AI & Machine Learning engineer with hands-on production experience in computer vision, edge deployment, and full-stack ML systems. Delivered a 94.1% accuracy multi-class species classifier on a 55,000+ image dataset at scale, built end-to-end inference APIs, and deployed models on resource-constrained edge hardware. Research paper currently under review.
NAME
Vishwas N
LOCATION
Bengaluru, IN
STATUS
Seeking Roles
FOCUS
CV · ML · Edge AI
STAGE
Final-Year Engineering
PUBLICATION
Paper Under Review
02 — CAPABILITIES

CONTROL
STACK

Programming

01
PYTHON (ADVANCED)JAVASCRIPTDART / FLUTTEREMBEDDED CSQL

AI / ML

02
PYTORCHTENSORFLOWKERASSCIKIT-LEARNRESNETVITBLIP-2LSTMU²-NETLAMAOPENCV

MLOps & APIs

03
FASTAPIFLASKREST APISMYSQLPOSTGRESQLMONGODBGEMINI APIGOOGLE COLAB

Edge & Embedded

04
RASPBERRY PIESP32ARDUINOLORAEMBEDDED CIOT SENSORS

Cloud & DevOps

05
AWS EC2CLOUDFLAREGITGITHUBLINUX

Full Stack / Mobile

06
FLUTTER (DART)REACT.JSHTML5CSS3

Cybersecurity

07
BEHAVIORAL MALWARE ANALYSISPROCMONVIRTUALBOX SANDBOXINGOS INTERNALS
03 — OPERATIONS

FIELD
OPERATIONS

4 ACTIVE DOSSIERS
◎ CASE_0012025
AgriSight
REAL-TIME PLANT DISEASE DETECTION
EDGE AI · COMPUTER VISION · CLOUD
  • Architected a hybrid edge-cloud inference system: ResNet fine-tuned in PyTorch, deployed on Raspberry Pi for sub-second local inference, with AWS EC2 handling retraining pipelines.
  • Implemented live camera streaming to the edge device for continuous crop disease monitoring without cloud dependency — critical for low-connectivity field environments.
  • Optimized model footprint for Pi deployment while maintaining classification accuracy across multiple plant disease categories under hardware constraints.
  • Demonstrated full MLOps cycle: data collection → training (EC2) → edge deployment → live inference → iterative retraining.
PYTORCHRESNETRASPBERRY PIAWS EC2OPENCV
◎ CASE_0022024 – 2025
All-Terrain Reconnaissance Robot
LONG-RANGE LORA TELEMETRY & AUTONOMY
EMBEDDED · IOT · ROBOTICS
  • Engineered a ruggedized autonomous reconnaissance robot with long-range LoRa wireless control and telemetry in GPS-denied environments.
  • Programmed motor control, sensor fusion (IR, ultrasonic) and navigation logic in Embedded C on Arduino/ESP32, achieving stable traverse across sand, gravel, and uneven terrain.
  • Implemented obstacle avoidance and real-time telemetry relay over a LoRa mesh network, enabling operator situational awareness from a remote dashboard.
ARDUINOEMBEDDED CLORAESP32SENSOR FUSION
◎ CASE_003AUG – DEC 2025
Polymorphic Malware Detection
BEHAVIORAL LSTM CLASSIFIER — PAPER UNDER REVIEW
CYBERSECURITY · AI RESEARCH
  • Designed a Windows 10 VirtualBox sandbox to safely capture behavioral profiles of polymorphic malware: API call sequences, file system events, and memory activity.
  • Implemented an LSTM sequence model classifying malware by runtime behavior rather than static signatures — resilient to obfuscation and polymorphic mutations.
  • Paper on hybrid detection using static analysis, function call graphs, and GAN/VAE generative augmentation currently under review for conference publication.
LSTMPYTHONPROCMONVIRTUALBOXGAN / VAE
◎ CASE_004APR – JUN 2025
Gemini Vision Annotation Pipeline
55,000+ IMAGE AUTO-ANNOTATION AT SCALE
COMPUTER VISION · AUTOMATION
  • Automated annotation pipeline using OpenCV for artifact removal and Gemini Vision API for bounding-box detection and square-crop generation across 55,000+ images with JSON audit logging.
  • Implemented pause/resume checkpoint logic in Colab, enabling fault-tolerant batch processing of large-scale datasets without data loss on interruption.
OPENCVGEMINI VISION APIPYTHONGOOGLE COLAB
04 — SERVICE RECORD

TIME
LINE

Nature's Ark
JAN 2026 – PRESENT
SOFTWARE DEVELOPMENT INTERN · BENGALURU, INDIA
  • Building an Urban Tree Map application (Flutter + Cloudflare) from early stage — owning dataset collection and structuring geospatial tree observation data for AI classification integration.
  • Designing data collection workflows and annotation pipelines to support training a species classification model on urban flora, bridging field capture with model-ready formats.
  • Contributing to backend architecture decisions for API design and data storage, ensuring scalability as the dataset and user base grow.
NextWealth Entrepreneurs Pvt. Ltd.
APR 2025 – NOV 2025
MACHINE LEARNING INTERN — COMPUTER VISION & AI SYSTEMS · BENGALURU, INDIA
  • Trained a ResNet-50 transfer learning classifier on 55,000+ annotated butterfly images across 1,046 species — achieving 94.1% top-1 accuracy on a highly imbalanced fine-grained dataset.
  • Engineered a multimodal BLIP-2 + Vision Transformer pipeline fusing image features, geolocation, and date metadata to produce automated species name + scientific classification.
  • Built end-to-end image preprocessing using U²-Net background removal, LaMa inpainting, and Gemini Vision API smart cropping to generate clean 512×512 training images at scale.
  • Solely engineered the iButterfly Count App (Flutter) — full mobile app for field biodiversity tracking with GPS tagging and structured observation logging. Published on Google Play Store.
  • Contributed model training, pipeline integration, and feature development for the Butterfly Explorer App, extending ResNet + BLIP-2 backend with discovery-mode species browsing. Published on Google Play Store.
05 — SECURE CHANNEL

OPEN
CHANNEL

Hiring for CV / ML / Edge AI? Working on something that needs a disciplined engineer? Send a transmission — I read every message.

✉ EMAIL
vishwasn1706@gmail.com
in LINKEDIN
vishwas-n
⌥ GITHUB
VishwasN1706
⚲ BASE
Bengaluru, IN
✉ TRANSMIT MESSAGE