Experienced in building compilers for deep learning accelerators.Good understanding and working experience of Deep Learning models (CNN) deployment and run time. Experienced with Tensorflow, Pytorch ONNX, MXNet and Keras frameworks.Experience with data analytics and model building (supervised and unsupervised). Experience in model miniaturization (reduce model footprint, low RAM use) methods like quantization(4 & 8bit) & pruning both from usage & architecture perspective.Tensorflow github contributor.Experience with tensorflow-serving, implemented Caffe models to serve using TFS.Apache TVM Committer, contribution in relay, ops, topi, rpc & various frontends.

MPLS-TE, RSVP, FRR (one to one & Facility), auto and manual bypass, Link & Node Protection, CSPF(Constrained SPF), PCE, PCE-SR(Segment Routing), Return Delegation. Prototype for egress node protection.

Have hands on experience on OpenFlow, OVSDB, OpenVSwitch(OVS), OpenStack etc. Involved in the development of Huawei SDN Controller (Agile Controller AC) based on OpenDayLight(ODL) and ONOS for Data Center(Use Java language). The current module which I'm working manages the switching and routing of VM traffic through VLAN/VXLAN.

Worked in SNMP, NETCONF, SSH, RADIUS, TACACS, FTP, TELNET, TFTP protocols for HUAWEI VRPV8 and FENIX Platform. These platform support multiple LS’s (Logical System), multiple VS ‘s (Virtual System), HA (High Availability), NSR (Non Stop Routing) and NSF (Non Stop Forwarding) features. Debugging problems in complex systems including, analyzing crashes, scaling and performance issues, memory leaks and corruption.